You are on page 1of 47

[Pick the date]

RANKO MATASOVI

A Grammatical Sketch of Albanian for students of Indo-European

Zagreb 2012

Ranko Matasovi University of Zagreb

A Grammatical Sketch of Albanian for students of Indo-European


1. Spelling
The modern orthography of Albanian, which will be used here, was adopted in 1908, at the congress of Monastir. It should be noted, however, that older linguistic books, such as Pokornys etymological dictionary, still use the antiquated transcription adapted by Gustav Meyer in the nineteenth century. Before that, Albanian used to be written in the so-called Elbasan or Todhri alphabet, which had developed from the cursive Greek alphabet. 1.1. Vowels <a, e, i, o, u> have more or less the Latin values (<e> is slightly more open than its Latin equivalent). <y> denotes a high front rounded vowel (German <>), and <> is a schwalike reduced vowel similar to French e muet. Final is not pronounced in Gheg dialects, which also have long and nasal vowels not reflected in writing. 1.2. Consonants There is little to be said about <p, t, k, b, d, g, s, m, n, v, z>. Stops are pronounced without aspiration, like in Romance. <q> and <gj> are palatal stops, like Sanskrit c and j. <th> and <dh> are voiceless and voiced interdental fricatives, respectively. <c> stands for the affricate /ts/, and <x> for /dz>, while <> is the voiceless affricate /t/, and <xh> is its voiced counterpart /d/. <sh> is //, and <zh> is //, while <h> is a laryngeal fricative like English h. <nj> is palatalized n (/n/), and <j> is a glide like English y. There are two different r-sounds, a flap <r>, and a trill <rr>. Similarly, there are two l-sounds, a velarized sound represented as <ll> and a regular <l>.

2. Chronology
Nothing is known about Albanians until the 11th century, when they are mentioned by Byzantine historians. The Albanian language is mentioned for the first time outside Albania, in the Croatian city of Dubrovnik, in the vicinity of which there appears to have been an Albanian community. In 1285, a man named Matthew recorded this sentence in the context of investigation of a robbery: Audivi unam vocem clamantem in monte in lingua Albanesca. Not long after that, Anonymi descriptio Europae Orientalis (1308) states that the Albanian language is not related to any of the neighboring languages, and thus confirms that lingua Albanesca is indeed the predecessor of modern Albanian: Habent enim Albani prefati linguam distinctam a Latinis, Grecis et Sclavis ita quod in nullo se inteligunt cum aliis nationibus.

Much has been written about the origin of the Albanian language. The most probable predecessor of Albanian was Illyrian, since much of the present-day Albania was inhabited by the Illyrians during the Antiquity, but the comparison of the two languages is impossible because almost nothing is known about Illyrian, despite the fact that two handbooks of that language have been published (by Hans Krahe and Anton Mayer). A lot of enthusiasm for Illyrian was lost after it had been discovered that the only purportedly Illyrian inscription, the Ring from Kelaja Delmas, was actually in Byzantine Greek. The inscription was read, from top to bottom, ANA OHQH ICER, and there were a lot of speculations about its meaning and grammar. However, a careful paleographic and philological examination showed that the actual reading is, from bottom to top, KE BOHQH ANA, where KE is the usual Byzantine abbreviation of the word Krie Lord (voc.). The meaning is thus O Lord, help Anna! in straightforward Byzantine Greek. Since that discovery, which almost destroyed Illyrology as a linguistic discipline, reasonable doubts have been expressed whether the Illyrians spoke a single language, or a group of more or less related languages roughly corresponding to the Italic complex. It is a priori less probable to assume that a single language was spoken in the whole Illyricum, from the river Arsia in Istria, to Epirus in Greece, when such a linguistic uniformity is found nowhere else in Europe before the Roman conquest. Moreover, the examination of personal names and toponyms from Illyricum shows that several onomastic areas can be distinguished, and these onomastic areas just might correspond to different languages spoken in ancient Illyricum. If Illyrians actually spoke several different languages, the question arises - from which Illyrian language did Albanian develop, and that question cannot be answered until new data are discovered. The earliest documents of the Albanian language stem from the 15th century. Besides a few almost unintelligible lines from the so-called Bellifortis Manuscript from 1405, the first sentence of Albanian we have is the baptism formula from 1462: Unte paghesont premenit Atit e birit et sperit senit I baptize you in the name of the Father, the Son and the Holy Ghost. 1 It is preserved in a work by Paulus Angelus, the Archbishop of Dyrrachium (Alb. Durrs). From 1483 we have one Albanian sentence in the Renaissance comedy Epirota by Tommaso de Mezzo (draburi to clofto goglie = trambur t kloft golja may your mouth tremble (from sickness)!), and from 1497 the earliest document of Albanian of some length, the glossary by a German traveler Von Harff. Von Harff was an adventurer who wrote a kind of tourist guide in which he collected valuable information about various European languages including Basque, Modern Greek and Albanian. He provided his reader with the native equivalents of such useful phrases as How much does this cost?, What is that, and Woman, I want to sleep with you. There are also a number of Venetian documents from 13th-15th centuries in which sevearal Albanian names are mentioned, as well as a few nouns. The earliest documents of Albanian literature sensu stricto are from the 16th and 17th century. These are religious works associated with the Counter-Reformation movement, and most of them are in the Gheg dialect. The first Albanian book is Meshari by Gjon Buzuku (1555, preserved in a single copy), and this was soon followed by works of Lek Matrnga (Catechismo di F. Ledesma tradotto da Luca Matranga, 1592, in Tosk), Pjetr Budi (Catechismo, 1618 and Spiegazioni della messa romana, 1621, Speculum confessionis tradotto dallo Specchio di confessione, 1612), Frang Bardhi (the author of the first Albanian dictionary Dizionario latino-epirotico,1635), Pjetr Bogdani (Cuneus
1

In the modern language, this would be Un t pagezoj pr emrit t atit e t birit e t shpirtit shenjt.

prophetarum de Christo salvatore mundi, 1685), and Nilo Catalano (Dizionario albaneseitaliano e italiano-albanese, 1694). Taken all together, very few books were published in Albanian in the 17th and the early 18th centuries. The first grammar of the language is Osservazioni grammaticali nella lingua albanese, published by Francesco Maria da Lecce in 1716, who also wrote a dictionary (in 1702) of some 13 000 entries. What we lack in early documents of Albanian, we gain by a close examination of the contemporary dialects. Albanian dialects are divided into Tosk (spoken chiefly in Albania south of the river Shkumbi) and Gheg (spoken in Northern Albania and in Kosovo, which is still part of Yugoslavia). The most salient difference between Gheg and Tosk is the so-called Tosk rhotacism: in Tosk, PAlb. *n > r word-internally and finally, whereas in Gheg n is preserved. Thus the ethnonym Albanian is Arbr in Tosk, and Arbn in Gheg. The modern literary language is based on the Tosk dialect of Southern Albania whose native speaker was Enver Hoxha, the former dictator of Albania. Outside Albania and Kosovo, there are very archaic Albanian dialects spoken in Greece (Sofiko, Salamis), Italy (Vaccarizzo), and Croatia (Arbanasi). The comparative and historical research of Albanian is also fortunate in that the Albanian vocabulary is loaded with loan-words from Latin, Greek, and various forms of South Slavic, some of which are very old 2. Since we can discover the sound changes that affected these loan-words, we are often able to reconstruct in great detail the shape of Proto-Albanian native words. With what has been said in mind, we adopt the following chronology: 1. Pre-Proto-Albanian (? -1st century B.C.). This is the period before the earliest contacts of Latin and Albanian. 1. Early Proto-Albanian (1st century B.C. -6th century). This is the period of intensive borrowing from Latin into Albanian, before the earliest contacts with the Slavs. 2. (Late) Proto-Albanian (6th century - 15th century). This is the period of intensive contacts of Albanians and Slavs. 3. Early Albanian (15th century - 1800). Roughly, this is the period of the earliest Albanian writings, as well as the period during which most of the Turkish loan-words entered the language. 4. (Modern) Albanian (1800 - present).

3. Historical Phonology
The phonological system of the standard language can be represented as follows: Vowels: i y e
2

u a o

In Meyers Etymological dictionary of Albanian, of 5140 keywords 1420 are Romance, 540 Slavonic, 1180 Turkish, 840 Modern Greek, and only 400 have a more or less reliable IE etymology. 730 words have no etymology whatsoever. During the past century, I would say that the number of words with IE etymology has risen, while some of Meyers Romance etymologies have been rejected, but the number of loan-words in Albanian is still disproportionately high.

The Gheg dialects also have nasalized vowels, e.g. Tosk hn moon vs. Gheg hn /ha/ Consonants: stops p t q k b d gj g fricatives f v th dh sh zh h affricates s c x xh

Resonants and glides: nasals m n nj ng laterals ll l rr r glide

3.1. Stress Early Proto-Albanian probably had a dynamic stress on the first syllable. Early Latin loan-words stressed on the penultimate lost the unstressed initial syllable(s), cf. L amicus >> Alb. mik friend, L impertor >> Alb. mbret king, L vicnus >> Alb. fqinj neighbour. In contemporary language the stress is regularly on the last syllable, but there are exceptions, cf. lle flower (but muz museum, jt life (but kt this), etc.

3.2. Vowels PIE *(H1)e > Alb. je in open syllables, cf. PIE *h1es(s)i > Alb. je you are, PIE *h1eptake (Hitt. e-ep-zi takes) > Alb. jep give. In closed syllables (before -CC-) the outcome is ja, cf. PIE *gwhermo- hot (G therms) > Alb. zjarm fire. Both of these changes must be dated after the Early PA, because they also affect Latin loan-words, cf. qiell sky << L caelu, fjal word << L fabella. Before a nasal followed by a fricative, Alb. e > i, cf. Alb. mish meat < PIE *mmso- (cf. OInd. mamsa-, with a long vowel, and OCS mso, L membrum < *memsrom, with short e). Before two consonants and followed by i, PAlb. *e (of whatever source) becomes i, cf. Alb. del he exits, but dilni you (pl.) exit, likewise heq draw but hiqni. PIE *o > Alb. a, cf. PIE *nokwt- > Alb. nat night, Alb. dark dinner, evening meal < PIE *dorkwo- (G drpon id.). Latin o is preserved, cf. L socius companion >> Alb. shok. Word-initial Latin o- receives a prosthetic v-, cf. Alb. vepr work << L opera, Alb. i varfr poor << L orphanus orphan (<< G orphans). PIE *a, to the extent that there was such a vowel in the proto-language, is represented as Alb. a, cf. PIE *kap-mi I hold (L capio, OHG haban have) > Alb. kam I have (note

the similar semantic development hold > have as in Germanic). Cf. also Alb. bath (pl.) broad beans, which seems identical to G phaks lentil, but the reconstructed form *bhako- contradicts the rules of PIE root structure, so it was probably borrowed from some substratum language. After velars, PAlb. *a > e (and the velars get palatalized), cf. L canem dog >> Alb. qen, calice chalice >> Alb. qelq, gallum cock >> gjel. One should also bear in mind that Alb. a (from PIE *o, *H2e) can be umlauted to e, especially in the plural, cf. Alb. n. sg. plak old man, pl. pleq. Sometimes forms with umlauted e are synchronically singulars, but originally plurals, so that Proto-Albanian *a vocalism should be reconstructed in the root, e. g., Alb. rreth circle < PAlb. pl. *rrathi (OInd. rath- chariot, L rota wheel). This process of substituting the plural form for the singular is quite frequent in Albanian, and is called singularization. PIE *e and *eH1 > Alb. o, cf. PIE *meh1(kwe) > Alb. mos prohibitive particle, PIE *ghesr hand (Hitt. kissar, G kher) > Alb. dor (via a secondary a-stem *ghesra), PIE *dheh1keh2 > Alb. dok custom, habit, G thke, PIE *leh1d- be tired (G lden, L lassus < *lh1d-to-) > Alb. lodhem am tired. PIE *o, *eh3, and *oh1 > Alb. e, cf. PIE *yeh3s- (G zvnh, Lith. jostas girt) > Alb. ngjesh I gird (n- is from a prefix *en-), PIE *nos we > Alb. ne. This change appears to be Late PAlb. since it affects Latin loan-words, cf. Alb. tmer fear << L timore, Alb. arsye reason << L ratione, pem fruit << L pmum, perhaps also her time << L hora (the word-initial h- in this word is dubious, since it was no longer pronounced in Latin by the time contacts with PAlb. were possible). Latin unaccented o and o are reflected as Alb. u-, cf. uroj wish, congratulate << L orare pray, urdhr order << L ordine, kuptoj understand << L computo compute. PIE *eh2 (*a) > Alb. o, cf. PIE *meh2ter mother > Alb. motr sister. The semantic shift is unusual, as Albanian is the only IE language in which the reflex of PIE *meh2ter means anything but mother. Alb. moll apple(-tree) can be from PIE *meh2lo-, parallel to G me lon but it is more probable that both the Albanian and the Greek words for , apple were borrowed from some unknown substratum. PIE *u is reflected as u, cf. Alb. nus daughter-in-law, originally bride < *nubhtyeh2 (cf. L nubo : nupta). It has been claimed that this is a loan-word from Latin (nuptiae wedding), but the meanings do not match very well. Word-initial u- is preserved in uri hunger < *un- (Gheg un) < *unh1- (zero-grade of *wenh1- > OHG wunscen wish, Skr. vnate desire). PIE *i remains, cf. Alb. dit day < *di-ti- (cf. OCS dn < *di-ni), i lig bad < *h3ligo- (G olgos small). *uH seems to give Alb. i, cf. Alb. mi < *muHs mouse, ti < *tuH you, tani now, if it is from *to-nu < *to-nuH (OCS nyn now), thi pig, if it is dissimilated from *ss < *suHs (Skr. s-). Perhaps *uH > Alb. u after *y, cf. *yuH- you (pl.) > Alb. ju (cf. Lith. jus, OInd. yuyam). However, the resemblance of Alb ju and PIE *yuH- could also be misleading,

since *y- is usually reflected as gj- in Albanian, so some scholars (Pedersen, Huld) prefer to connect Alb. ju to PIE *we- seen in L vos, OCS vy. It is difficult to find original instances of *iH in Albanian. The most often quoted example, Alb. pi drink, probably involves laryngeal metathesis (PIE *peh3i- ~ zero-grade *ph3i- > *pih3-, cf. OCS piti drink), and is not altogether reliable. Albanian y comes regularly from the sequence wi (and is spelled as ui in Early Albanian, e. g. in Buzukus works), cf. qytet city < L civitatem (Acc.). Alb. y can also be the result of umlaut, cf. fryt product < *fruti << L (pl.) fructus, as well as the regular outcome of L u in loan-words, cf. pyll forest < *padule < L paludem swamp (acc.), brym icicles < bruma. Alb. is a reduced vowel, which usually occurs in non-syncopated pretonic, and nonapocopated post-tonic syllables, cf. Alb. kndoj sing << L cantare, trboj enrage << turbare. It also occurs in the first syllable of some native words, regularly before a nasal followed by a consonant, cf. dhmb tooth < *gombhos (G gmphos comb), ndrr dream < *h3nro- (cf. G nar). 3.3. Diphthongs The outcome of PIE diphthongs is another source of dispute in the historical phonology of Albanian. The following seem to be the most reliable etymologies: PIE *ey > Alb. i, cf. dimr (Gh. dimn) winter < PIE *gheymon (G ce?mwn). PIE *oy > Alb. e, cf. shteg path < *stoygho- (OCS stza id. < *stigheh2), be faith < *bhoydeh2 (L fides < *bhid-), ver (Gheg ven) wine < *woyno- (G oi nos, L vinum). The outcome of *h2ey (*ay) is uncertain. If keq bad can be connected to L caecus blind, Goth. haihs, OIr. cech, it would appear that *ay > e; similarly, edh goat, when compared to G ax, aigs points to *h2eyg- (but this word is likely to be a borrowing from a non-IE source). PIE *ew > Alb. e, cf. hedh throw < *skewd- (OE sceotan). At least word-initially, *h2ew and *h2ow seem to yield Alb. ve-, cf. Alb. vesh ear, which is from PIE *h2ews- ear (L auris) Alb. ve egg, which must be somehow related to L ovum, OCS j-aj (PIE *h2owyom, a vrddhi to *h2owis bird), and Alb. vet self, if it is related to G auts id. (from ?*h2ewto-). This rule seems to be contradicted by Alb. ag, agim dawn, but its relation to G aug, OCS jug south is far from clear. The outcome of *h2ew, *h2ow, and *ow word-internally is uncertain. If ter bull is not a loan-word from Latin, as Haarmann (1972) thinks, but a native word corresponding to L taurus, G taurs, OIr. tarb, then it would appear that *-aw- > -e-. In Latin loan-words L au > Alb. a, cf. Alb. pak few << L paucu, Alb. gaz joy << L gaudiu, ar gold << L aurum.

There are a few diphthongs in Albanian, whose origin is very dark, e. g., ye, which occurs mostly in plurals, and appears to be due to some kind of compensatory lengthening, e. g., dyer, pl. of der door < *dhwor-, or krye head. There is also the socalled broken vowel, ua, which is represented as uo in the early writers (Buzuku, Budi). In Gheg dialects this diphthong is either monophthongized to u, or pronounced as ue (NE Gheg). It occurs in some irregular plurals (duar, pl. to dor hand) and in some difficult forms (e. g. grua woman, buall waterbuffalo bull).

3.4. Auslautgesetze The development of vowels in final syllables in Albanian is difficult to ascertain. Here are some rules that seem to apply quite regularly: PIE *o and *e are lost, cf. n. sg. of o-stems, e. g. *ulkwos wolf > Alb. ujk (this is a dialectal variant of PIE *wlkwos wolf, OInd. vrkas, that Albanian seems to share with Celtic, cf. OIr. olc bad and names in Ulc-, Olc-). The development of *e can be seen in the n. pl. of the consonant stems, e. g. PIE *nokwtes nights > Alb. net. PIE *eh2 > *a is retained as - in open syllables, but as -e in closed syllables, cf. n. sg. and pl. of a-stems, e. g., drit light (n. sg.) < *drkta, drite lights < *drktas. PIE *i seems to be lost in open syllables, but preserved as - in closed syllables, cf. (in open syllable) *esmi > Alb. jam I am, *h2elbhi barley > Alb. elb (G lphi), (in closed syllable) *nokwtis > Alb. nat night, perhaps also *penkwtis five > Alb. pes. PIE *o (and *-oH) was probably narrowed to *u, like in Proto-Celtic, and then shared the outcome of PIE *-uH > -i, cf. PIE *dwoH two > PAlb. *dwu > *dwi > Alb. dy. By positing this change it is easy to account for the 1st sg. present ending in -i of verbs such as di I know. After vowels, this i was reduced to -j, cf. kuptoj I understand < *kuptao << L computare. Final *e seems to be reduced to , at least in closed syllables, cf. motr sister < PIE *meh2ter. Final *oy first developed to *e, just like any other *oy, and then it fell off, often causing umlaut in the preceding syllable, cf. n. pl. of the o-stems in *-oy (an old pronominal ending also seen in L lup, Lith. vilkai volwes), e. g. miq friends, n. pl. of mik << L amicus.

3.5. Laryngeals in Albanian? The idea that at least one of the laryngeals was preserved in Albanian is due largely to Eric Hamp (1965). Hamp thought that PIE *h4, which in his view colors *e to *a, but disappears in Hittite, survives in Albanian as h. Indeed, there are some words in Albanian that contain an unexplained h-, but the etymologies offered by Hamp are rather poor (for a thorough criticism, see lberg 1972). The only universally acknowledged source for Albanian h-, however, is from PIE *sk-, cf. hije shadow < *skeh2i- (G ski<,

OInd. chay-), hn moon < *skandna (L candor), hudhr garlic, either related to or borrowed from G skrodon. Reconstructing the fourth PIE laryngeal to account for obscure Albanian words like heq draw, push, hirr whey, hund nose, and hekur iron, seems to me out of the question. The only case where Alb. h- occurs in a word with clear IE correspondences with initial laryngeal is Alb. herdh testicles, which is certainly cognate with G rkhis, OIr. uirge, and Lith. erilas, arilas stallion, and in this case the reconstructed laryngeal is PIE *h3 or *h1. The PIE word for testicle should be reconstructed as *h3orghis or *h1orghis, and Alb. -e- is probably due to singularization of an original plural, so we should posit an original PAlb. n. sg. *hardh(). Although there do not seem to be any obvious counter-examples 3 to a sound law *h1/3V- > Alb. hV-, one word is certainly not a solid basis for far-fetched conclusions. One should also mention that Alb. rreth circle, which is certainly related to L rota wheel, OInd. ratha- chariot (PIE *(H)roth2o-) has an unexplained th rather than *t, which might point to a development *tH > th in Albanian, but in view of the lack of further examples this remains a speculation. Before resonants laryngeals disappear word-initially, cf. ndr dream < *h3nr- (cf. G nar, Arm. anurj), njeri man < *h2ner- (G anr), perhaps emr (Gh emn) name if from *h3nomn (Gr. noma, Skr. nma, etc.). The reflexes of laryngeals after syllabic resonants are discussed below.

3.6. Resonants Resonants are generally preserved, unless one takes into account the Tosk rhotacism, by which *-n- > -r-. PIE *m > m, cf. Alb. mal hill < PIE *mlh3dh- (OE molda forehead, G blothrs high reaching), Alb. mjel I milk < PIE *h2melg- (G amlgo, L mulgeo, Eng. milk), mjalt honey < PIE *melit- (G mli, mlitos; cf. also Alb. blet bee < *mlVt-). PIE *n > n, cf. Alb. nnt 9 < *newn-ti-, PIE *newn (L novem, etc.), Alb. nye knot, if related to L nodus. In Tosk, -n- > -r-, cf. Alb. ver wine (Gheg ven) < *woyno- (L vnum). In Latin loanwords, n- remains, cf. Alb. numr number << L numerus. PIE *l > l, cf. Alb. lig bad < PIE *h3ligo- (G olgos few). Word-initial l- in Latin loanwords is represented as ll- before back vowels (including a), but as l- before front vowels, cf. lepur rabbit << L lepore, ligje law << L lege, lir free << L liber, but llaft << laude, llar laurel << L lauru. PIE *r > r, cf. Alb. rjep catch < *h1rep- (L rapio, G erptomai), Alb. dru tree < *drun-, PIE *doru-/dru- (OInd. dru-). Initial *wr- yields Alb. rr-, cf. Alb. rrnj root < *wrVd-n(OIr. frn, G rza < *wridy). Word-initial r is usually represented as rr-, cf. Alb. rreth circle < PIE *(H)rth2- (German Rad wheel, Skr. ratha- chariot). This is also the case with Latin loanwords, cf. rret net
3

Alb. asht bone is more likely than not from PIE *Hh2osth1-, like G oston, L os, ossis, and Hitt. hastai-.

<< L rete, rrot wheel << L rota, rrobull oak << L robur. If Latin r was preceded by a vowel, it is represented as Alb. r-, cf. ran sand << L arena. PIE *w > v, cf. Alb. vej weave < *webhnyo (PIE *webh- > Germ. weben, etc.), PIE *widheweh2 widow > Alb. e ve (OInd. vidhava, L vidua, OIr. fedb, etc.). In Latin loanwords, v- is preserved, cf. Alb. va ford << L vadum. PIE *y > Alb. gj, cf. Alb. n-gjesh gird (OInd. yas- girt, OCS po-jas), Alb. gjuaj hunt, if related to Germ. jagen. Alb. gj- is the only assured reflex of PIE *y, and the change is confirmed by Latin loan-words, cf. Alb. gjykoj judge << L iudicare. Some think that there is another possibility, namely PIE *y > Alb. z, cf. Alb. zej bake and G zo boil, or Alb. i zi, fem. e zez black (fem.), as if from *yeh3dyeh2-, cf. Lith. jodas black. The usual source of Alb. z- is *dy-, both in native words and Latin loan-words, cf. zan fairy << L Diana.

3.7. Syllabic resonants The traditional view is that PIE *r, *l > Alb. ri, li, whereas PIE *n, *m > Alb. an, am. These reflexes would be similar to the reflexes in Celtic. However, if a laryngeal follows, *rH > Alb. ar, *lH > Alb. al. Here are some examples: PIE *R: Alb. drith grains < PIE *ghrd- (L hordeum barley, NHG Gerste), Alb. krimb (Gheg krymb, with regular change of i > y before labials) worm < PIE *krmi- (OInd. krmi-, Lith. kirms); the relation of this word to L vermis and Eng. worm is unexplained. PIE *RH Alb. i bardh white < PIE *bhrHgo-, Alb. i par first < *prh3wo- (OInd. purva-, OCS prv), Alb. mal hill < PIE *mlHdho- (OInd. murdhan- head), Alb. val wave < PIE *wlHnis (Lith. vilns id.; the presence of laryngeal in this root is shown by the intonation of Latv. velt walken). There are, however, problems with the traditional view. Namely, in several cases PIE *r appears to yield Alb. ar where a following laryngeal cannot be reconstructed, cf. Alb. ar bear < PAlb. *ar (- is a productive suffix) < PIE *h2rtko- (Hitt. hartagga-, G rktos, L ursus, OIr. art), Alb. ardhur, pple. of vij come < *h1rgh- (G rkhomai, OIr. regaid will go). Unless one wants to blame this irregularity on the initial laryngeal, we can claim that word-initially PIE *r > Alb. ar.The only problem with this view is Alb. rit I rear ritem I am growing, which Meyer relates to OInd. rdhnti, rdhti prospers, succeeds. The t instead of d in the Albanian word cannot destroy this etymology, since stops are devoiced word-finally in several dialects, but perhaps the rather vague correspondence of meanings makes it not so attractive. Mayrhofer relates OInd. rdhti to G lthomai grow, succeed, which seems plausible, and in which case Alb. rit has no etymology.

10

We would also expect word-initial *l to give Alb. al-, but there are no examples. If Alb. liqe, Tosk liqr lake is the same word as L lacus, G lkkos, OCS loky pond (PIE n. sg. *lokus, gen. sg. *lkwes), it would appear that *l- > Alb. li-. A special problem is Alb. grun wheat, which should be somehow related to L granum, Goth. kaurn < PIE *grHno-. We could posit a change *ar > ru after velars, but then Alb. gardh fence, hedge must be considered a (?Germanic) loan-word (cf. Goth. gards house, yard, Lith. gardas enclosure, fortified town). Similarly, if shurr urine is from *skrneh2 (see below), its vocalism would be irregular, but a similar development of *r > ur after a velar is attested in gur stone < PIE *gwrH- (OCS gora mountain, Skr. gir- mountain).

3.8. Consonants 3.8.1. Stops PIE voiceless stops are preserved, cf. Alb. pjek bake < PIE *pekwoh2 (OInd. pacati, L coquo), Alb. pi drink < PIE *peh3i- (G pn, OInd. pbati), Alb. ti you < PIE *tu(H) (OInd. tuvam, L tu, OCS ty), Alb. katr 4 < PIE *kwetwores (OInd. catvaras, G tttares; it has also been claimed that Alb. katr is a loan-word from L quattuor, but this is improbable), Alb. nduk pull < *dewk- (L dc). Voiceless stops are voiced after nasals, cf. prind parent << L parente, mbrnda inside << per intu, nnd nine < *newn-ti-. This change is late, since Early Albanian texts still have -nt-, -mp-, e. g. Von Harffs dictionary has nente for 9. The development of PIE voiced and aspirated stops is unclear. It is certain that *bh > b, cf. Alb. bjer bring (imperative) < PIE *bhere (OInd. bhra, G phre, L fer). One cannot find any convincing examples of reflexes of PIE *b in Albanian. Hulds derivation of Alb. det sea from PIE *dhewboto- (cf. OE deop deep, OCS dubok, Lith. dubs) is possible, but tells us little. Proto-Albanian *b and *d, from whatever source, are sometimes lost word-internally, even in loan-words, cf. Alb. e ve widow < *widheweh2 (L vidua), Alb. fe religion << L fides, fjal word << L fabella, luaj play<< L ludere, mjek physician << L medicus. It appears certain that reflexes of *g and *gh are different in Albanian, in which case aspirated and voiced stops were still distinct in PAlb. The regular reflex of *g is dh, whereas *gh is reflected as d: PIE *g Alb. dhmb tooth < PIE *gombhos comb (G gmphos, Lith ambas). Note the same change in meaning as in OCS zb tooth. Alb. dhndhrr son in law < *genh1- (Lith. ntas, OInd. jat relative). Alb. i bardh white < PIE *bhrHgo- (OInd. bhurj- shining). Alb. mbledh gather < *en-legoh2 (G lgo speak, L lego read).

11

Alb. madh great < PIE *megh2 (OInd. mahi, G mga, L mag-nus). PIE *gh Alb. dor hand < PIE *ghesr (Hitt. kiar, G kher, Toch. A tsar) Alb. dimr winter < PIE *gheymon (G khemon, L hiems, OCS zima). Alb. drith grains < PIE *ghrd- (L hordeum, OHG gersta, perhaps also Arm. gari). Alb. ardhur, pple. of vij come < PIE *h1rgh- (G rkhomai I shall go = OIr. regaid). This is not a counter-example to *gh > d, since the change *d > dh /r_ is regular and also affects Latin loan-words, cf. urdhr order << L ordine. Since the changes *g > dh and *gh > d seem assured, one would expect PIE *d to give Alb. dh, and *dh to give Alb. d. However, both *d and *dh give Alb. d almost without exception: PIE *d Alb. dy two < PIE *duwoH (OInd. duva, G do, etc.). Alb. dru tree < *drun- (PIE *doru / *dru- > G dru, g. doratos < *dorwntos). Alb. dit day < *ditV- (OCS dn). Alb. dem small bull < *domo- (OInd. damya- bull OIr. dam < PIE *dom(H)-). Alb. drit light < *drkta (OInd. dr- see, G drkomai). This etymology is a bit weak on semantic grounds. Alb. djerr wasteland < *derno- (if related to G dero peel, OCS der). Alb. deg twig, branch < *dwogho- (cf. E twig, NHG Zweig). PIE *dh Alb. djeg burn < PIE *dhegwh- (OInd. dahati, Lith. degti). Alb. ndez kindle < *endhogwheyo- is a causative built to the same root, PIE *dhogwheyoh2 > L foveo). Alb. der door < *dhworu- (G thra, L fores, Lith. dvras < PIE *dhwor / *dhurs). Alb. dje yesterday < PIE *dhghes (G khths, L here < *hesi, OEng. geostra). Alb. dele sheep, if related to G thelus feminine, L felo suck (PIE *dheh1i-l-). A very small number of words seem to have Alb. dh for PIE *d. Since this is the expected outcome on structural grounds, the question arises whether these forms are relics, or do they show the results of a late development of d > dh. Here are the

12

problematic forms: Alb. dhjet 10 < PIE *dekm (G dka, L decem, etc.), Alb. dhi hegoat, if it is related to NHG Ziege, Alb. lodhem am tired < PIE *leh1d- (G lden), and Alb. dhash I gave, aorist to jep give (PIE *deh3- give > G ddomi, L do, dare, etc.). It is easiest to explain these forms if one assumes that *-d- > -dh- (before the merger of voiced and aspirated stops). In dhjet ten dh- is generalized from compound numbers, and in dhash I gave from forms with preverbs. Alb. dhi can have dh rather than *d by analogy with edh goat, which is probably the same word as OInd. aj-, OPr. wosee goat (PIE *h2ogo-; the Albanian form implies a lengthened grade, PIE *h2ogo-). There are no instances of Alb. dh for PIE *dh. Alb. dhe earth should be derived not from PIE *dheghom (Hitt. tekan, G khthn, etc.) but rather related to G (Hom.) gai a earth, which is also of uncertain origin.

3.9. PIE Plain velars in Albanian? It has been claimed that the difference between the three PIE series of gutturals is preserved in Albanian before front vowels. This thesis, sometimes referred to as Pedersens law, is often contested, but still supported by the majority of Albanologists (e. g. Hamp, Huld, and lberg). In examining this view, one should bear in mind that it seems certain that there were at least two palatalizations in Albanian: the first palatalization, whereby labiovelars were palatalized to s and z before front vowels and *y, and the second palatalization, whereby all the remaining velars (*k and *g) were palatalized to q and gj, in the same environment. PIE palatalized velars are affected by neither palatalization (they yield Alb. th, d, dh, cf. Alb. thom I say < *kensmi, cf. OInd. ams- praise, L censeo reckon). It may be that th yielded f before a consonant, if Alb. nfle sleep is from *nthle < *n-kloye- (cf. G kln recline).

3.10. The first palatalization PIE *kweloh2 I turn > Alb. sjell bring, fetch, turn, G (Hom.) plo turn, move, become, PIE *meh1-kwe > Alb. mos (prohibitive particle), cf. G m-te, PIE *kwih1 how > Alb. si, PIE *gwhermos heat > Alb. zjarr, zjerm fire, G therms hot Arm. jermn fever, PIE *ghwen- sound (OCS zvon) > Alb. z voice. Perhaps this process can account for Alb. sy eye, if it should be derived from an old dual *h3kwih1 (G Hom. sse, OCS oi eyes). Its -y- rather than *i must somehow be analogical.

3.11. The second palatalization This process affects many Latin loan-words, e. g., qershi cherry << cerasiu, qind 100 < centu, gjind people << gente, shgjet arrow << sagitta. Since Alb. gj has several other sources (PIE *y, *s), its occurence in native words is not alarming, but the occurence of q in native words is, since this sound can be neither from PIE *k (it would have changed to th) nor from *kw (it would have been palatalized to s). Yet there are quite a few instances of q in Albanian that cannot be considered Latin loan-words, e. g. qoj wake, rouse, which seems to correspond to, but cannot be derived from, L cieo set in motion (cf. also G Hom. ko go, and kino set in motion, perhaps also OInd. cyavate). All these forms can be explained by positing a PIE root *keh1i-, with zero-

13

grade (and laryngeal metathesis) *kih1-. Alb. qeth cut has been compared to Lith. kert id., but this equation has with some justification been put into doubt, since one would not expect *r to be lost before th (see abej 1972: 144). On the other hand, the etymological connection of Alb. gjej (gjenj in dialects) find and G khandno, L praehendo (praeda spoils < *prai-heda), E get still seems fine to me 4. Moreover, PIE *k seems to be preserved before PIE *e, which later changed to o in Albanian, cf. koh time which can be superimposed on OCS as moment. Note, however, that the Albanian form actually implies a feminine noun *kesa, or *keska, the Slavic form is from a masculine *kesos, or *keskos, and still in OPr. we have neuter kisman time. However few they are, instances of PIE plain velars in Albanian should be taken seriously, since there does not seem to be an easy way to get rid of them. Although many aspects of Albanian historical phonology remain unclear, this is by itself not an argument against Pedersens law. In order to refute it, one would have to devise a scenario in which 1. All of the Albanian native words with k, g, q, gj are from PIE words with labiovelars, or desatemized palatals (on which see below). 2. In all such words the reflexes of PIE labiovelars preceded back vowels during the period of the first palatalization, and 3., those back vowels were subsequently - analogically or otherwise changed to front vowels. Since no such scenario seems to be forthcoming, the problem of the validity of Pedersens law must remain open.

3.12. Desatemization in Albanian The development of PIE gutturals is further complicated by the fact that PIE palatals seem to be regularly depalatalized before resonants in Albanian 5. Again, there are few examples of this change, but no counter-examples, which makes the rule convincing, e. g., PIE *smekru chin (OInd. maru-, Lith. smkras, OIr. smech, perhaps also Hitt. zamankur) > Alb. mjekr id., PIE *klu- hear (OInd. ru-, G klo) > Alb. quhem am called, PIE *greh2u- old woman (G graus) > Alb. grua woman, wife (but note that there is a different etymology of this word, relating it to G gyn woman), PIE *gonu- / *gnew- knee (OInd. jnu, G gnu, L genu, Goth. kniu) > Alb. gju, def. gjuni, gjuri.

3.13. PIE *s PIE *s gives gj- word-initially, cf. PIE *supnos sleep (G hupns, OCS sn) > Alb. gjum, PIE *sokwos sap, juice (G ops, OCS sok) > Alb. gjak blood. The change, which can be represented as *s > * > * > gj, must have been very early, since Latin borrowings regularly have Alb. sh- for L s-, cf. Alb. shok friend, comrade << L socius, Alb. shum much << L summum. The development of *s word-medially is unclear. It
However, another explanation is possible, though difficult, if one starts from *ghednyo, which would regularly give *gjanj, *gjaj in Albanian. The problem is whether *-d- was lost before or after *e > ja in the closed syllable, and whether this in turn preceded or followed the second palatalization. The e-vocalism of the attested form should then somehow be analogical. 5 A similar rule applies in Balto-Slavic, but there back vowels must follow the resonant. See R. Matasovi, Poredbenopovijesna gramatika hrvatskoga jezika, Zagreb 2008.
4

14

has been claimed that the outcome is h, as in koh time, if from *keso-, but both Alb. koh and OCS as moment can also be derived from PIE *kesko-, and the change *sk > Alb. h is regular anyway. A case can be made for the development *s > sh, cf. Alb. gen.-dat.-abl. pl. (indefinite) -sh, if it is from PIE *-oysu (OInd. -eu), and mish meat < *mso- < *memso- (OCS mso). In any case, Latin loan-words have -sh- wordinternally for Latin -s-, cf. Alb. rrshin << L resina, Alb. dshiroj want << L desiderare, Alb. kafsh thing << L causa. In sequences *s...s, the first sibilant is dissimilated to th, cf. Alb. thi pig < PIE *suHs (L sus, G hus, Alb. thaj to dry < *sh2ewso- (OCS sux dry). It has been claimed that PIE *s- > sh- before unstressed vowels, cf. shtat 7 < *septm,(G hept), shi rain < *suH-(G hei it rains), but this is uncertain. A particular problem is presented by Alb. shurr urine, which has been related to Hitt. ehur id., but which is more likely to be from something like *skrneh2, and thus related to G skr, skats excrement (a r/n stem). Word-initial *sw- seems to yield Alb. d-, at least when stressed, cf. Alb. diell sun < *swel- (Skr. svar-), dirs sweat < *swid-r- (Skr. svedate sweats).

3.14. Consonant clusters In clusters involving two stops, the first stop is lost very early, cf. nat night < *nokwt-, shtat 7 < *septm, tet 8 < *(H)okto. In Latin loanwords, *pt and *kt yield ft, cf. L luctor, luctri fight >> Alb. luftoj, L presbiter (<< G presbteros) priest > *prepter >> Alb. prift. Clusters involving resonants are also simplified: *gn > nj, cf. njoh know (a person) < *gneh3V- (L co-gnosco, G gignsko). *kl > q (kl is preserved in some dialects and in Buzuku), cf. quhem I am called < *klusk- (OInd. ru- hear), L ecclesia church > Alb. qish. Alb. q is also the result of Latin c before front vowels, e.g. in qytet town < civittem, faqe face < L facis. *gl > gj, cf. gju knee < *gnu- (OInd. jnu, G gnu), Lat. glandula gland > Alb. gjndr, also gjuh if it is a loanword from G glssa. Early documents (Buzuku) and some dialects still have gl. *dl > gj, perhaps in gjat long, if it is from something like *dlHgho- (-t is a late suffix), cf. OInd. dirgh-. *dy > z, cf. Lat. Diana > Alb. Zan (a fairy), Lat. gaudium > Alb. gaz joy. *b(h)n and *pn > *mn > m, cf. gjum sleep < *supno- (G hpnos), am river-bed < *abno- (L amnis river, OIr. aub < PIE *h2eph3on / gen. *h2eph3nes). *rd > rdh, cf. pjerdh break wind < *perdoh2 (OInd. pardate, G prdomai). Also in loanwords from Latin, cf. verdh yellow < L viride.

15

*lk > jk, cf. Alb. ujk wolf < *ulko- < *wlkwo- (Lith. vilkas, G lkos); also in Latin loanwords, e.g. fajkua falcon << L falconem. *lw > ll, cf. Alb. gjall living < *solwo- whole, safe (OInd. sarva-, G hlos whole). *rn > rr, cf. Alb. verr alder < *werno- (Gaul. verna alder, OIr. fern), also in loan-words, Alb. furr oven << L furnax. The developments of clusters involving *s are very complex. Here are some examples: PIE *ks > Alb. sh, cf. *seksti- 6 > Alb. gjasht, *ekstos, *eghstos outside (G Locr. ekhths, OIr. acht but) > Alb. jasht. PIE *sp- > Alb. f, cf. *spora seed, that which is sown > Alb. far (G spros), perhaps fruth measles, if from something like *sprkto- (OInd. sprni- speckled, L spurcus dirty). It has also been claimed that *sp- > shp-, cf. shpend bird, which can be derived from *spetno-, with s-mobile (cf. *petno- > OIr. n bird). PIE *sl > Alb. ll, cf. PIE *h2us-li- spark (OEng. ysle) > Alb. yll star. 3. 15. Other consonants Albanian consonants , xh, zh can mostly be derived from earlier clusters, but they are also very frequent in loan-words, especially from Turkish, but also from Slavic, cf. Alb. xhep pocket << Tur. cep id., xhezve small coffee pot << Tur. cezve id., oban shepherd << Tur. oban id. izme boot << Tur. izme id., udi wonder << Serb. udo id.. In some cases, Alb. is from earlier * (written sh in Modern Albanian, e.g. moj estimate < *shmoj << Lat. aestim). In oj (aor. uaj) lift, take to, wake up, which is abstracted from earlier n onj (Buzuku, with the prefix n-), we may have the root found in cy drive and qoj wake, rouse if one assumes the development from *nq- to n-. The consonants c and x are rather rare and occur in expressive words and those with obscure etymology, e.g. xix spark, caporre cow with long horns. In some cases cappears to be from *ds-, e.g. ca some besides disa some (by syncope), or cep edge besides dialectal step edge (metathesis). Likewise, Alb. cy (2-3sg. pres. cyt) drive, impel may be related to L cieo move and G aor. kie started to move if we assume that the Proto-Alb. form contained the prefix *d- (as in dbor snow besides bor id.) and the dental suffix *-t- (from the past participle stem in *-to-?), i.e. that the development was from PIE *kyu-to- to Proto-Alb. *d-syu-t- and to Alb. cy. The consonant x sometimes comes from Romance palatalized velars, e.g. in lexoj read (cf. Italian leggere). The consonant f is common in Latin loan-words, in which it develops from v, k, p before dentals (kafsh animal << L causa, kofsh hip << L coxa, aft capable << L aptus).

16

4. Morphology
4.1. Nouns and adjectives

4.1.1. Noun Noun has five cases (nominative, accusative, genitive, dative, and ablative) and two numbers (singular and plural). It also has a very complicated system of suffixed definite articles, so that for each noun has definite and indefinite declensional paradigms. Albanian nouns are either masculine or feminine, with a residual class of neuters denoting mostly mass nouns (e.g. uj water, mish meat. The PIE declensional types were merged as o-stems (masculine and neuter), -stems (feminine), and i-stems (masculine and feminine) in Proto-Albanian. Here are examples of indefinite paradigms: o-stems; mik friend sg. NAcc. shok GDAbl. shoku a-stems; vajz girl sg. NAcc. vajz GDAbl. vajzaje pl. vajza vajzave, vajzash pl. NAcc. shok GDAbl. shokve, shoksh

Since final syllables were lost in PAlb., the PIE case-endings disappeared, with the possible exception of Loc. pl. *-su, which is preserved as (G-D-Abl. pl.) -sh. The ending GDAbl. sg. o-stems is u after velars and i elsewhere (e.g. dem-i, GDAbl. of dem young bull); this ending has been derived from the locative sg. of the o-stems *-ey, although the more common PIE ending was *-oy (OCS vlc, L sg. of vlk wolf). Nominative plural is formed very differently for different classes of nouns, cf. mik friend ~ miq, zog bird ~ zogj, fshatar villager ~ fshatar, vajz girl ~ vajza, kujtim memory ~ kujtime, bri horn ~ brir, lum river ~ lumenj, bir son ~ bij. Many nouns have completely irregular plurals, cf. djal boy ~ djem, der door ~ dyer, ka ox ~ qe, thes sack ~ thas, grua woman ~ gra, nat night ~ net, bir son ~ bij. As can be seen from the examples, there are three types of plural formation, and in some nouns they are combined: a) addition of the suffix (bri horn ~ brir); b) change of the stem final consonant (mik friend ~ miq), c) Ablaut (grua woman ~ gra). Some nouns have the borrowed Turkish pl. ending ller, -llre (< Turk. ler/-lar, depending on vowel harmony), e.g. baba father pl. baballar.

17

The ablaut in the formation of plural in many nouns was caused by the adding of the suffix *-oy > , which was subsequently lost, after causing palatalization of the preceding velar (e.g. plak old man < *plako-, pl. pleq < *plak). In some cases, this plural formation was influenced by other factors, e.g. in gjarpr snake, pl. gjarpinj we have an original n-stem, in which *r changed to n in the sg. (Tosk rhotacism). Some nouns (both and o-stems) have a special vocative form, e.g. neno mother, burro husband; these were probably influenced by the Slavic vocatives of the -stems (OCS eno, voc. sg. of ena woman).

4.1.2. Adjectives The adjectives have masculine and feminine forms; the masculine and the feminine forms are often distinguished only by different forms of the article, e.g. i mir good (m.) vs. e mir good (f.); some adjectives have different forms for the m. and f. in the singular and in the plural, e.g. i ndryshm different (m.) vs. e ndryshme (f.), pl. t ndryshm (m.) t ndryshme (f.). The gradation of adjectives is analytic, as in other Balkan languages, e.g. i madh big, m i madh se bigger than, i mir good, shum i mir best.

4.2. Verbs The history of the Albanian verbal system is largely unexplored. There are very few things about it that can be stated with certainty. Albanian verb has two numbers (sg. and pl.) and distinguishes three persons. It forms several tenses synthetically (present, past (aorist), and imperfect), and has many periphrastic tenses (future, future perfect, perfect, and pluperfect). It also has a rather developed system of moods (indicative, imperative, subjunctive, optative, and admirative).

4.2.1. The present tense In principle, there are two major conjugations, thematic (punoj work, vendos decide) and athematic (jam am, kam have). Examples of the present tense of athematic verbs: jam am sg. jam je sht di know sg. pl. di dim di dini di din jap give ha eat sg. pl. sg. pl. jap jepim ha ham jep jepni ha hani jep japin ha han ka have sg. pl. kam kemi ke keni ka kan

pl. jemi jeni jan

The athematic endings more or less straightforwardly go back to the PIE athematic present: PIE *h1esmi (Skr. smi) > Alb. jam, *h1(e)smes > jemi), but the 3rd sg. sht (dial. sht) contains an epenthetic *n (*ensti), presumably under the influence of 3pl.

18

jan < *(e)senti. The origin of the 2pl. ending is disputed. Some linguists see it as aglutinated from the original *-te- (Gr. te) and the particle *nuH now (Gr. ny n ). Here are three thematic verbs in the present tense: punoj work sg. punoj punon punon laj wash sg. laj lan lan vendos decide, place sg. pl. vendos vendosim vendos vensosni vendos vendosin

pl. punojm punoni punojn

pl. lajm lani lan

The origin of several endings is quite unclear. In the 1sg., the athematic ending is from PIE *-mi (Gr. d-d-mi I give, etc.), while the thematic ending presumably comes from *-o-m. In Buzukus works, the 1. sg. ending (-onj) still retains the nasal. Several verbs form nasalized present stems, e.g. bind to convince < *bheydh- (Gr. pethomai), tund to move, stir < *tud- (Lat. tundo strike). In some cases these presents are preserved as Alb. present stems in j- < *nj, e.g. gjej find < ghed-n- (Gr. khandn, Lat. prae-hendo, etc.). Present stems formed with the suffix *-sk- > Alb. h- include fsheh hide < *skep-sk- (Gr. skp cover) and shoh see < PIE *skw-sk- (Skr. sacate follows). The progressive present form is formed by putting the particle po before the present tense form of the verb: po punoj I am working. There are many irregular verbs. Here are the paradigms of eci walk and dua want: sg. 1. eci 2. ecn 3. ecn pl. ecim ecni ecin sg. dua do do pl. duam doni duan

4.2.2. The past tense (aorist) Examples of the past tense (aorist): kam have sg. 1. pata 2. pate 3. pati pl. patm patt patn punoj work sg. punova punove punoi pl. punuam punuat punuan jap give sg. dhash dhe dha pl. dham dhat dhan

The past tense of kam is formed from a suppletive stem pat- which can perhaps be related to PIE *poti- master, lord (cf. L possum can < *potis sum). The past tense of jap is also suppletive: it goes back to PIE *deh3- give, which acquired the sigmatic aorist (dhash < *dh3-s-m, dha < *dh3-s-t, etc.), cf. also rri sits aor. ndenji, sheh sees aor. pa, vjen comes aor. erdhi and ha eats aor.hngri.

19

The endings partially correspond to those of the present (e.g. in the 1. and 3pl.). The 1sg. ending a might go back to PIE athematic (syllabic) *-m. In the 2pl., the aorist preserves the PIE ending *-te, which was lost in the present. Several verbs have the reflexes of the lengthened -grade in the past, e.g. djeg burn aor. dogja < *dhgwh- (cf. Skr. dhati burns), bredh jump aor. brodha < *bhrdh- (cf. OCS bred wade), hedh throw aor. hodha. The lengthened -grade was generalized from the PIE sigmatic aorists. In a number of cases analogical lenghtened -aorists were formed to roots in *a, e.g. marr take aor. morra, dal go out aor. dola. There are probable reflexes of the PIE sigmatic aorist in Albanian. These include, e.g., the aorist of the verb jap give, which is suppletive: 1dhash < *dh3-s-m dhame < *dh3-(s)-me 2 dhe < *deh3-s-s dhate < *dh3-(s)-te 3 dha < *dh3-s-t dhan < *dh3-(s)-nt Other Alb. formations that can be derived from PIE sigmatic aorist include thash said (aor. of thom) < *kns-s- (Lat. censeo think) and rash fell (aor. of bie) < *(H)rew-s(Lat. ruo rush). Note that this aorist formation is rarer in Old Albanian authors such as Buzuku, so in many cases it represents an analogical extension. Some authors think that the ending sh is not actually from the sigmatic aorist, but rather from the agglutinated imperfect of the verb to be (3 sg. ishte). The asigmatic aorists are mostly from Late PIE thematic aorist formations, e.g. erdha came (suppletive aor. to vij come) < *h1ergh- (Gr. rkhomai go, OIr. regaid will go, also vesh put on clothes aor. vesha < PIE *wes- dress oneself (Hitt. weta, Skr. vste), puth kiss aor. putha, laps weary aor. lapsa, rras squeeze aor. rrasa). Many Albanian verbs form the aorist with the suffix v-, which has been compared with the suffix found in Lat. v-perfects (amv loved, docu taught, etc.), but it is uncertain if these formations are really related. In many cases, *-w- was actually part of the root, so we are dealing with original root aorists. These Albanian aorists include lava (aor. of laj wash) < *leh3w- (Gr. lo), dava (aor. of daj divide) < *deh2y- (Gr. daomai), theva (aor. of thyej break), etc. Several aorist forms are of obscure origin. These include the aorists in t- (e.g. ryj enter aor. ryjta, gjaj resemble aor. gjajta, luaj move, shake aor. lojta, di know aor. dita) and the aorists in r- < *-n- (by Tosk rhotacism), e.g. hy< enter aor hyra, bj do aor. bra.

4.2.3. The imperfect The imperfect was originally formed from the present stem, but in roots with the vowel -e- it was remodeled with the umlauted wovel i-, cf. the imperfect of sht is and punoj work: sg. 1. isha 2. ishe pl. ishim ishit sg. punoja punoje pl. punonim punonit

20

3. ishte

ishin

punonte

punonin

Here are the paradigms of pjek cook and vras kill, which have vowel alternations: 1. piqja 2. piqje 3. piqte piqnim piqnit piqnin vritja vritje vriste vritnim vritnit vritnin

As can be seen, the endings are similar to those of the (asigmatic) aorist (1sg. ja, 2sg. je, 3. sg. te, 1pl. nim, 2pl. nit, 3pl. nin). In Buzukus works, the nasal is still retained in 1. and 2. sg. (-nj, -nje). In Gheg, the element sh- from the verb to be was re-interpreted as an imperfect suffix and it spread throughout the paradigm (expect in the 3sg.), cf. Gheg punojsha I was working, punojshe you were working, etc.

4.2.4. The perfect and the pluperfect. The perfect is formed analytically with the present of the verb ka have, as in the Romance languages; likewise, the pluperfect is formed with the imperfect of ka. These are the paradigms of punoj work: Perfect: 1. kam punuar kemi punuar 2. ke punuar keni punuar 3. ka punuar kan punuar Pluperfect: kisha punuar kishe punuar kishte punuar kisim punuar kishit punuar kishin punuar

The pluperfect corresponds quite well to the English pluperfect. There is, in general, concordance of tenses: Marku mendoi sa shum kishte msu-ar pr dy jav Mark.nom.sg. think.1sg.aor. how much has.3sg.ipf. learn-pple in two week Mark thought how much he had learned in two weeks

4.2.5. The future The future is formed analytically, according to the same pattern as in the Balkan languages, i.e. with the 3rd sg. present of the verb di want and the subjunctive of the inflected verb (cf. Serbian (Torlak) e da radim I will work, e da radi you will work, etc.). Here is the future of punoj work: 1. do t punoj 2. do t punosh 3. do t punoj do t punojm do t punoni do t punojn

21

In the Gheg dialect, there is also the Romance-type future tense formed with the present tense of kam to be and the (Gheg) infinitive: kam me shkrua I will work, ka me shkrua he will work, etc.

4.2.6. The mediopassive The present and the imperfect of the mediopassive are formed synthetically. The other tenses are formed analytically with the clitic u < *swom, e.g. aorist u lava I washed myself. Here are the paradigms of laj wash: Present sg. 1. lahem 2. lahesh 3. lahet Imperfect sg. lahesha laheshe lahej

pl. lahemi laheni lahen

pl. laheshim laheshi laheshin

The origin of the suffix e- in the mediopassive is unknown. The endings may be partially derived from the PIE middle endings, e.g. the 2sg. present ending sh may be from PIE *-so(y) (Skr. bhara-se you carry yourself), and the 3sg. ending t may be from *-to(y) (Skr. bharate). The other present mediopassive endings were obviously influenced by the active paradigm. The origin of the 2pl. imperfect ending i is mysterious. The present subjunctive mediopassive is formed with t and the present mediopassive, and the future with do want and the subjunctive (e.g. do t lahem I will wash myself). The perfect mediopassive is formed with the present indicative of the verb jam to be and the participle, cf. the paradigm of the verb hap open: sg. 1. jam hapur 2. je hapur 3. shte hapur pl. jemi hapur jeni hapur jan hapur

The aorist mediopassive is formed with the particle u which precedes the active aorist of the inflected verb (except in the 3sg. where there is no ending in the mediopassive): sg. 1. u hapa 2. u hape 3. u hap pl. u hapm u hapt u hapn

The imperfect mediopassive has its own set of endings, which are essentially the imperfect active endings added to the suffix esh-: sg. 1. hapesha 2. hapeshe 3. hapeshe, hapej pl. hapeshim hapeshit hapeshin

22

The pluperfect mediopassive is formed with the imperfect of the verb jam to be and the participle, e.g. isha hapur I had been opened, ishe hapur you had been opened, ishte hapur he/she had been opened, etc.

4.2.7. The moods Most of the moods in Albanian have disputed origins. Some are clearly recent formations, like the admirative, which is derived with the present of the verb ka appended to the stem of the participle: a) The admirative of punoj work: 1. punuakam punuakemi 2. punuake punuakeni 3. punuaka punuakan The admirative expresses surprise, admiration, doubt, but also evidentiality: U, sa e bukur qenka kjo vajz Oh, how beautiful.fsg. be.admirative.3sg. this.f. girl.nom.sg. Oh, how beautiful is this girl! Thuhet se Gjoni qenka one.says that Gj. be.admirative.3sg They say that Gjoni is the culprit fajtori the.culprit.nom.sg.

The admirative is rare in Old Albanian (16-17th centuries), and it is not attested in the dialects of Greece and Italy. It appears that in Buzuku it functions as a focus marker. b) The optative appears to have recent origin; it is formed with the suffix fsh- (-sh- after consonants, --after n-) and the (mostly) aorist/subjunctive endings: punoj work sg. 1. punofsha 2. punofsh 3. punoft hap open sg. hapsha hapsh hapt jam be sg. qofsha qofsh qoft

pl. punofshim punofshi(t) punofshin

pl. hapshim hapshi(t) hapshin

pl. qofshim qofshi(t) qofshin

The suffix fsh- has been explained as the analogical extension of the 2nd sg. ending -sh-, which was originally added to the aorist stem in v- (e.g. kerkov-a sought), with the regular change of vsh > fsh. In the 3rd person sg. -fsht > -ft. The used to express wish and desire, but it also occurs in some subordinate clauses including conditionals. Q-ofsh i gzuar be-2sg.opt. rejoice(participle) May you be happy

23

Edhe n fjet-sha, zemr-n sma v dot njer-i n gjum even if sleep-1sg.opt heart-A.sg.def.neg.1sg.D.A. put particle man-Nsg.def. to sleep Even if I sleep, no-one (lit. no man) will put my heart to sleep N shk-ofsha n Angli, do t t lajmr-oj if go-1sg.opt. to England will you.A.clitic part. inform-1sg.fut. If I go to England, I will let you know c) The subjunctive has synthetic forms preceded by the particle t. Here is the present subjunctive of punoj (there are also analytically formed perfect, imperfect, and pluperfect paradigms): 1. t punoj 2. t punosh 3. t punoj t punojm t punoni t punojn

The endings of the subjunctive are mostly the same as in the present active, but in the 2sg. the mediopassive ending is used. In Buzukus language, the 2. and 3. sg. endings still have the nasal (2. sg. onjsh, 3. sg. onj), and the 1. sg. ending is 0 rather than -oj. The subjunctive is the chief mark of subordination, as it usually occurs in subordinate clauses, e.g. with verbs of intention: Dshir-on t pun-oj n kopsht want-3sg.pres. to work-3sg.subj. in garden He wants to work in the garden Shpres-oj se tash do t je-t i hareshm hope-1sg. that now want.3sg to be-3sg.subj. m. cheerful.m I hope he will be cheerful now It is also used with verbs such as filloj begin and the modals such as mund can and duhet must: Ai filloi ti krye-j detyr-at he begin.3sg.aor. to.acc.pl.clitic. finish.3sg.subj. duty-acc.pl. He began to finish his duties A mund t m lshoni nj viz particle can to I.dat.sg. let.2pl.subj. a visa.Acc.sg. Can you issue a visa to me? far mund t blesh atje? what can buy.2sg.subj. here What can you buy here / what can one buy here? A duhet t paguaj dogan? particle must pay.1sg.subj. customs.duties

24

Do I have to pay customs duties? In some cases the use of the subjunctive is optional (it can be replaced by the indicative, e.g. in temporal clauses introduced by do dit, kur Every day, when. The subjunctive may also be used as an adhortative, or to express a wish in the main clause (in this function it is parallel with the optative): T bj be Kalosh Cami! do.3sg.subj. trust K. C. May Kalosh Cami swear (it)! T shkojm Lets go! d) The conditional is formed with the 3rd sg. of di want and the particle t (as in the future) followed by the imperfect of the inflected verb: 1. do t punoja 2. do t punoje 3. do t punonte do t punonim do t punonit do t punonin

The perfect conditional may be formed with the imperfect of the auxiliary ka have and the participle (do t kisha punuar I would have worked, do t kishe punuar you would have worked, etc.). e) The imperative has just the forms of the 2 sg. (ji be!, puno work!, shkruaj write!, fol say, ki have!, hap open!) and the 2 pl. (jini, punoni, shkruani, flisni, hapni). In the other persons an analytical construction with the imperative of la let is used (cf. the English construction with let), e.g. le t jet may he be, le t ken may they have, le t thomi let us say. When the imperative has a pronominal object, it is cliticized to the imperative form and written together: hap-e open it, shkruaj-e write it, merr-e take it. In the plural, the object clitic precedes the 2pl. ending, thus we have pij-e-ni drink (pl.) it, merr-e-ni take (pl.) it, etc.

4.2.8. The infinite forms of the verb Like most Balkan languages, Albanian is poor when it comes to infinite forms of the verb. It has an all-purpose participle formed from the aorist stem (punuar from punoj work, lar from laj wash, fol-ur from flas speak, kthyer from kthey return, dal from dal go out, qe-n from jam be) and an infinitive formed from the participle (pr t punuar). There are also the present and the past gerundive (present duke punuar, past duke pas punuar). The most common form of the participle/infinitive shows the operation of Tosk rhotacism (cf. Old Gheg lan washed corresponding to Tosk and standard lar). In consonant clusters rhotacism did not operate, so we have thn said, qen been, dhn given (the nature of the cluster is difficult to ascertain in each individual instance).

25

The Gheg dialect has the infinitive fomr formed with the preposition me with and the participle (me shkrua to write). It is used in many constructions in which Tosk (and the standard language) use the subjunctive in subordinated clauses. The gerundive is formed analytically from the participle and the particle/adverb duke while, e.g. duke lar while washing.

4.3. Pronouns a) Here are the forms of personal pronouns in the Nominative: sg. 1. un 2. ti pl. na ju

The origin of 1sg. un is unknown; it has been derived from the reflexive u (< *swom) and a particle, perhaps related to G na. The form of the 2sg. is from PIE *tuH (OCS ty), and the 1pl. pronoun na is from PIE *no- (Lat. ns). The 2pl. pronoun is often thought to be cognate with Lith. jus, Skr. yyam (PIE *yuH-), but PIE *y- usually gives Alb. *gj- (see above), so it is more likely that the Alb. pronoun goes back to PIE *wes-/*us- (L vs, G hymei s) with some kind of prothetic j-. The personal pronouns are inflected; the forms of the 1sg. are: A/D mua (clitic m), Abl. meje; 2sg. A/D neve (clitic na), Abl. nesh; 1 pl. A/D ty (clitic t), 2 pl. A/D juve (clitic ju). The suppletion in the 1 sg. is inherited from PIE, cf. Gr. Acc. em, me, Dat. emo, moi. b) In the third person, the demonstrative ai m. ajo f. is used as the personal pronoun. It is formed from the pronominal stem *ey- (Lat. s, Skr. ayam, etc.). It is inflected according to the following table: m. sg. N A D Abl. ai at, t atij, tij atij ajo at asaj, saj (clitic i) asaj, asoje f.

pl. N A D Abl. ata ata atyre asish, atyre, atyreve ato ato atyre asosh, atyre, atyreve

The origin of most forms is unclear.

26

The personal pronouns also have clitic forms which distinguish case (Dative and Accusative) and number, but not gender:

1sg. A D m m 1pl. A D na na

2sg. t t 2pl. ju ju

3sg. m.f. e i 3pl. m.f. i u

Albanian, like most Balkan languages, has clitic doubling, which means that direct and indirect objects are coded twice: once with a full NP, and again with a clitic which agrees with it in case, person, and number: Un e ble-va nj libr I Asg.clitic buy-aor.1sg. a book.Asg.indef. I bought a book Un i shkro-va nn-s I Dsg.clitic write-aor.1sg. I wrote to my mother

mother-Dsg.def

Note also the following construction: Mua m plqen kyo vajz-a me.D me.D(clitic) like-3sg.pres. this.fem.sg. girl-nom.sg.def. I like this girl (cf. Italian mi piace questa ragazza). If a verb has both a direct and an indirect object, the dative and the accusative clitics merge into a single phonological word, e.g. m + e = ma, t + e = ta, I + e = ia, na + e = na e, ju + e = jua, u + e = ua, m + i = mi, t + i = ti, i + i = ia, na + i = na i, ju + i = jua, u + i = ua: Ma jep librin He gives me the book, jua jep librin He gives you the book, Ua jep librat He givest them the books, etc. c) The proximal demonstrative pronoun is ky m., kyo f., presumably from the PIE interrogative stem *kwo- (Lat. quod, Skr. kas who, OCS kto). It appears that the interrogative kush who contains the PIE interrogative stem *kwu- (Skr. kuh where) agglutinated with the demonstrative stem *so- (Skr. sa- m. this). In the Acc. sg. we have k < *kwom. Alb. () what may be from PIE *kwi- (Lat. quid what). This stem is also incorporated into ka what The pronoun tjetr another is from t jetr, where jetr < *H1e-tero- (OCS jeter). The reflexive pronoun vet oneself has been connected with

27

G auts < PIE *h2ewto-. It is inflected and freely combined with prepositions in reflexive constructions: Skam gj me vete neg-I.have thing with self I dont have anything on me (lit. with myself) Shiko vete-n look.imperative self-Acc.sg.def. Look at yourself! d) The possessive pronouns (used also as the genitives of the personal pronouns) form a complex system; they are inflected for case and gender. For illustration, here is the paradigm of the 1st person sg. possessive pronoun: m. N G-D A Abl. sg. im tim tim tim pl. e mi t mi e mi t mi sg. ime sime time sime f. pl. e mia t mia e mia t mia

e) The relative pronoun q (from the PIE interrogative/relative stem *kwo-) is indeclinable. It introduces the relative clause in which the grammatical relation of the relativized NP must be indicated by a resumptive dative or accusative clitic pronoun: Njeri-u q e kam par man-def.Nsg. that Acc.sg.clitic have seen The man I saw Grua-ja q I jap nj dhurat woman-def.Nsg. that Dsg.clitic give.aor.1sg. a gift The woman I gave a gift to There is also the inflected relative pronoun built from the stem cil- (of disputed origin): m sg. i cili i t cilit t cilit t cilin f. sg. e cila i t cils t cils t ciln m. pl. t cilt i t cilve t cilve t cilt f. pl. t cilat i t cilave t cilave t cilat

Nom. Gen. Dat. Acc.

It is more or less synonymous with the particle q, but its use does not require the resumptive pronoun in the relative clause. Moreover, it can be combined with the prepositions to express that the relativized noun functions as an oblique in the relative clause: Miq-t e mi me t cil-t kam ardhur ktu friend-Nom.pl.def. clitic my with who-Acc.pl. I.have come here

28

My friends with who I have come here

4.4. The article Albanian has two sets of definite article forms, the preposed article i (m.), e (f.) and the postposed (clitic) article i /u (after velars) m. and a f., cf. zog m. bird def. zogu, gur stone m. def. gur-i. Both the definite and the indefinite forms of the article are inflected for case and number. The origin of the forms is disputed, but it is likely that the ultimate source is the demonstrative (the stem *ey- of Lat. is, ea, id may lie behind the masculine form in the nominative sg.). The numeral nj one serves as the indefinite article, e.g. nj gur a stone. The postposed definite article is written together with the preceding noun, and the whole compound is inflected as the definite NP: gur m. stone sg. Nom. guri Acc. gurin G (i) gurit D gurit Abl. gurit pl. NVA G D Abl. fjal f. word

fjala fjaln (i) fjals fjals fjals

gurt (i) gurvet gurvet gurvet

fjalt (i) fjalvet fjalvet fjalvet

When definite nouns are modified by (postposed) adjectives, the adjectives have to be preceded by the definite articles, creating, as it were, agreement in definiteness: guri i bardh 'white stone: *guri bardh (gur m. stone is in the nom. sg.def.) gurit t bardh 'of the white stone': *gurit bardh (gur is in the gen. sg. def.) lulja e bukur 'beautiful flower': *lulja bukur (lule flower f. is in the nom. sg. def.) The connector used in these phrases is also inflected; the forms are: i (m. sg. Nom.) < *ey-, t (Acc. sg. and pl. m. and f., GDAbl. m. and f.) < *tom, s (Gen. sg. f.), and e (f. sg.). The complex pattern of inflection of NPs can be illustrated with the following examples: m. sg. 'good boy' indefinite N djal i mir A djal t mir

definite djali i mir djalin e mir

29

G i/e djali t mir D djali t mir Abl. djali t mir f. sg. 'good girl' indefinite N vajz e mir A vajz t mir G i/e vajze t mir D vajze t mir Abl. vajze t mir m. pl. indefinite N djem t mir A djem t mir G i/e djemve t mir D djemve t mir Abl. djemsh t mir f. pl. indefinite N vajza t mira A vajza t mira G i/e vajzave t mira D vajzave t mira Abl. vajzash t mira

i/e djalit t mir djalit t mir djalit t mir

definite vajza e mir vajzn e mir vajzs s mir vajzs s mir vajzs s mir

definite djemt e mir djemt e mir i/e djemve t mire djemve t mir djemve t mir

definite vajzat e mira vajzat e mira i/e vajzave t mira vajzave t mira vajzave t mira

With some (rather formal) terms for blood relatives the article has possessive function, e.g. i ati his father, i biri his son, i vllai his brother, e ma his mother, e motra his sister, etc.

4.5. The numerals The cardinal numerals from one to ten are derived directly from PIE: nj < *h3oyno- (Lat. nus) dy < *dwoh1 (Lat. duo) tre < *treyes (Lat. trs) katr < *kwetwores (Lat. quattuor, Gr. tttares) pes < *penkwe (Lat. qunque, Gr. pnte) gjasht < *seks(-ti-) (Lat. sex) shtat < *septm(-ti-) (Lat. septem) tet < *h3ektw(-ti-) (Lat. oct) nant < *newn(-ti-) (Lat. novem) dhjat < *dekm(-ti-) (Lat. decem).

30

As in most Balkan languages, the numerals from 11 to 19 are construed with the preposition meaning on (Alb. mb), cf. OCS jedin na deste 11: njmbdhjet 11, dymbdhjet 12, etc. Twenty is (nj-)zet, which cannot formally be connected to PIE *(d)wiH-dkmtih2 (L vigint, etc.), so its etymology is a mystery. The numerals from 30 to 90 are simple compounds: tridhjet 30, pesdhjet 50, etc., and 100 is qind << L centum. The ordinals are derived from the cardinal numerals with the suffix t < *-to, e.g. dyt second; Alb. par first is from PIE *prh3o- (Gr. prtos, OCS prv). They follow the noun they modify.

4.6. The Prepositions Only a few Albanian prepositions have reliable PIE etymologies: mbi on < *h2mbhi (G amph, OIr. imb-, L ambi-, Skr. abhi-). pr for, about < *per(i) (L per, G per) n in, on < *en(i) (L in, OCS v, OIr. in) Most prepositions take the ablative case (afr near, bri beside, drejt towards, gjat during, jasht out of, mes between, prfund underneath, lart up, larg far from, ans along, buz by, close to, etc.). A few take the nominative (nga from, towards, te(k) at, to), or the accusative (deri, gjer until, up to, mbi on, me with, by, ndaj near, m on, n in, on, pa without, q, qysh since, pr for).

4.7. The Conjunctions and the subordinators Here are the commonest coordinating conjunctions: dhe, e, edhe and as as neither nor a, apo, ose or po, porse, ndrsa but nse if ndonse although The subordinating conjunctions (subordinators) introduce subordinate clauses: I. Subordinators se and q introduce subordinate clauses with epistemic verbs: Atij iu duk se ajo matej ti thoshte dika to.him to.him appear.3sg.aor. that she wanted to.him say.3sg.ipf. something It appeared to him that she wanted to tell him something Nuk pretendohet q t ket lexuar Makiavelin not state.pass.pres.3sg. that have.3sg.subj. read Macchiavelli.Acc.sg. It is not stated (implied) that he had read Macchiavelli

31

II. Temporal clauses are introduced, e.g., by kur when and pasi after: Kur ai psritet shum her, athere kjo sht vij when it repeat.3sg.pass.pres. many time then it is line When it (scil. a silly thing) is repeated many times, then it is the (party) line III. Causal clauses are introduced, e.g., by sepse, nga q because Ato duan t arrijn shpejt, sepse ato jan t lodhura they want.3pl.aor. to arrive.3pl. quickly because they are art. tired.Nom.pl. They wanted to arrive quickly, because they were tired IV. Comparative clauses are introduced, e.g., by sikur as if: Besniku e mori aparatin fotografik n duar sikur tishte nj relike B. Acc.clitic take.aor.3sg. camera.Acc.sg. in hands as.if be.3sg.ipf. a relic Besnik took the camera in (his) hands as if it were a relic

32

4. The lexicon
In the preceding pages we have quoted numerous Albanian words with PIE etymologies, as well as a lot of Latin loanwords. It should be noted, however, that many sAlbanian words have no generally accepted etymology. For all we know, they could be loanwords from some non-IE substratum, e.g. zemr heart, vlla brother, hekur iron, ri new, -zet twenty, mund can, zog bird, helm poison, ur bridge, vajz (dial. varz) girl, i vogl small, i rnd heavy, rri sit, etc. Besides the Latin loanwords, the Albanian lexicon contains loanwords from several other languages. Early (mostly Byzantine) Greek loanwords include dhaskal teacher << G dskalos, krevet bed << G krebbti, sos to end << G aor. ssa I arrived, trndafil rose << G triantphyllon, drom road << G drmos, nisi island << G ns, etc. An earlier stratum of loans include words borrowed from Ancient Greek, often from Doric, e.g. mokr millstone << G makhan, drapr sickle << G drpanon, perhaps llr lower arm << G lna. There are a few Germanic loanwords which may have entered the language from Balkan Gothic, e.g. gardh garden << Goth. gards, fat husband << Goth. fas, shkum foam (cf. German Schaum). Early Slavic loanwords are, e.g., balt mud (Cf. OCS blato, Croat. blato), zakon custom (OCS zakon law), bised a talk, discussion (OCS besda), dalt chisel (Croat. dlijeto), grosh beans (Croat. graak green beans), nevoj need (OCS nevolja trouble), begati richness (OCS bogat rich), trup body, corpse (OCS trup), porosit request (OCS poriti), pusull note (OCS posl), grusht fist (OCS grst), vidr otter (Croat. vidra), etc. Some of these loanwords entered the language before some early Slavic sound changes in the 8th century, e.g. the liquid metathesis (balt, dalt), the change of *u to (pusull, grusht), etc. Turkish loanwords are present in all languages of the Balkan area, and Albanian is no exception, cf., e.g. oban cowherd (Tur. oban), budalla idiot (Tur. budalla), asqer soldier (Tur. asker), boj colour (Tur. boja), hoxha moslem priest (Tur. hoca), qejf pleasure (Tur. keyif), haber news (Tur. haber), etc. Of particular interest are words attested in Albanian and Romanian, where no Romance (or Latin) proto-form can be posited, e.g. Alb. neprk serpent vs. Rom. nprc, Alb. shtrung a shed for sheep vs. Rom. strung, Alb. boll a big serpent, a dragon vs. Rom. balaur big snake (cf. also Croa. blavor a lizard), brez girdle vs. Rom. bru, Alb. i bukur pretty vs. Rom. a se bucura to rejoice, Alb. vatr hearth vs. Rom. vatr (cf. also Croat. vatra fire), Alb. vjedhull badger vs. Rom. viezure, Alb. i bardh white vs. Rom. barz swan, Alb. avull smoke, fog vs. Rom. abur(e), Alb. sorr crow vs. Rom. coiar, etc. The origin of these words is uncertain. They are most likely to be prehistoric loans from Albanian (or some closely related, but extinct ancient Balkan language) in Romanian.

33

SAMPLE TEXTS
1. Ismail Kadare: Generali i ushtris s vdekur The General of the Dead Army (p. 1) The following extract is from the best known novel of the greatest contemporary Albanian writer, Ismail Kadare (b. 1936). The novel is about an Italian general who comes to Albania after the 2nd World War to conduct the exhumation, and arrange for the repatriation of the remains of Italian soldiers. The General of the Dead Army is a literary monument of the Stalinist socialist realism with Albanian flavor. In the flashbacks from the war, Albanian partisans slaughter Germans and Italians by batallions (though Germans never had more than one division in Albania), and post-war Tirana is represented as a modern metropolis with car traffic, night clubs, and foreign embassies, although those things first appeared there only a couple decades ago. Nevertheless, the novel is quite readable, and some passages are very well written. Mbi tokn e huaj binte shi dhe dbore przier bashk. Sqota kishte qullur betonin e pists s aeroportit, ndrtesat, rojat. Ajo lagte fushn dhe brigjet dhe shklqente mbi asfaltin e zi t xhades. Sikur t mos ishte fillimi i vjeshts, do njeriu tjetr, prve gjeneralit t porsaardhur, do ti dukej ky shi monoton nj koincidenc e trishtuar. Ai po vinte ne Shqipri nga nj shtet i huaj pr trheqjen e eshtrave t ushtarve t vrar ktu n luftn e fundit botrore. Bisedimet midis dy qeverive kishin filluar q n pranver, por kontratat perfundimtare u nnshkruam vetm n fund t gushtit, taman n kohn kur filluan vransirat e para. Pra, ishte vjesht dhe shiu kishte kohn e tij. Gjenerali e dinte kt. Englished: Rain and snow were falling simultaneously on the foreign land. The mixture of rain and snow had soaked the concrete runway of the airport, the buildings, and the guards. The snow-storm covered the plain and the hills, and it made the black asphalt of the road shine. As it still wasnt the beginning of autumn, to any other man, but to the general who had just arrived, it would appear that this monotonous rain was a sad coincidence. He had come to Albania from a foreign country to search for the remains of the soldiers killed here in the last world war. Negotiations between the two governments had begun already in spring, but the final contracts were signed only by the end of August, just at the time when the first grey days begin. Yes, it was autumn, and it was the rainy season. The general knew that. Etymological comments: mbi on, about: preposition, related to OInd. abhi, G mphi, OIr. (preverb) imb-, probably from PIE *H2mbhi. tok earth has no etymology. i huaj foreign is from something like *ksenyo-, or *skenyo-. The first possibility seems more probable in light of the Greek word for foreigner, ksenos. binte is an aorist of bie hit, fall < PIE *bheryoh2 (L ferio wound, Lith. barti). It is homophonous with bie carry, which is from PIE *bheroh2 carry. shi has been compared to G hei it rains, Toch. B swese rain, but this cannot be true, since PIE *s- gives gj- in Albanian. One could always posit an ad hoc rule PIE *s- >Alb. sh- in monosyllables, but this would be the only example of it.

34

dhe and, was originally (an emphatic?) particle. It cannot be related to L hoc (Hamp), or huic (Huld), since PIE *gh- would have given Alb. d-. The proto-form must have been PIE *go, so it can be related to G -ge, unless this latter particle is identical with OCS e, which is from *ge. In any case, PIE seems to have had a bulk of very similar syntactic particles, *ge, *ge, and *ghe (cf. OInd. ha). In the contemporary language, dhe is usually combined with the prefix e-, so and is edhe. Alb. e is probably from L et and (conjunctions can be borrowed, cf., e. g., Latv. un and from Germ. und). dbore snow has no satisfactory etymology. The all-purpose PIE verb *bher- carry was often invoked, but it is difficult to see both the semantic and the formal connection. przier mixed contains the common prefix pr, but is otherwise unclear to me. bashk together Etym? sqot f. rain and snow, snowstorm has no etymology. It is a rare word, not found in some dictionaries. kishte is the 3sg. imperfect of kam have. The present stem is from PIE (athematic) *kapmi hold (cf. L capio, Goth. haban have), but the imperfect stem is difficult to account for. Perhaps analogy with ishte was (from sht) played a role, but ishte also has unexplained i-. qullur drenched, wet, soaking. Used as a participle with the imperfect of kam it forms the plusquamperfect I, thus kishte qullur had soaked beton, pist, aeroport are all modern borrowings, with obvious meanings. E and s are connecting particles that Albanian uses to bind head nouns to following dependent genitives. ndertes building is a verbal noun from ndertoj buildt, which has no etymology. roj guardian. Etymology? ajo she, and ai he are usually derived from a particle a- (perhaps from the PIE pronominal stem *e-/*o-), and the PIE pronominal stem *so- (masc.), *seH2- (fem.). This is supported by the plurals ata (masc.), ato (fem.), because the PIE demonstrative showed suppletism of *s- (nominative sg.) and *t- (other cases and numbers). PIE *s is claimed to have disappeared in the juncture, and j is a hiatus-filler, thus we should suppose PIE *o-seh2 > PAlb. *a-so > a-o > ajo. I cannot say I am fully convinced of this, but there is no better explanation lagte is the imperfect of lag to wet, moisten. Perhaps can be compared to OIr. lobaid putrescit, lobur weak < PCelt. *lob-ro- < PIE *logw-. fush plain has no etymology. breg, pl. brigje hill, embankment is a Slavic loan-word, cf. Russ. breg bank (of the river), Croat. brijeg hill. shklqente is the imperfect of shklqej to brighten, a verb with no etymology. asfalt requires no explanation. i zi, fem. e zz black has been related to Lith. jodas black by abej. This etymology depends, however, on his theory that PIE *y sometimes gives z in Albanian, but this is far from established. t is an (inflected) particle used for connecting the noun with a dependent genitive. xhade road is a loan-word from Turkish (Tur. cadde id.). sikur if, as if. mos is a negation, related to OInd. ma, G m, Toch. AB ma, all from the PIE (prohibitive) negation *meh1. The Albanian form should be derived from *meh1kwe, where *-kwe is a conjunction (OInd. -ca, L -que, etc.). fillim beginning is a verbal noun derived from filloj begin. Etymology? vjesht fall, autumn has no etymology. do every, each. 35

njeri, def. njeriu man is identical to OInd. nr-, G anr man, PIE *H2ner, acc. *H2ner-m, gen. *H2nros. The vocalism of the Albanian word shows that the shape of the acc. sg. was generalized. -i is a productive suffix (cf. ari bear < *H2rtkos, Hitt. hartagga). tjetr other, another. Dialects show that the original form had j-, e. g. jetri in the dialect of Falconara. Standard Albanian tjetr probably owes its t- to the particle t (crasis). PAlb. *yetro- other can be compared to Umb. etro-, OCS jeter other. gjeneral does not require an etymological explanation; gjenerali is the definite form. porsaardhur is compounded from porsa- just now and ardhur. Ardhur is a (suppletive) participle of vij come, and it should be derived from PIE *h1rgh- (cf. G rkhomai, OIr. regaid shall go). As a noun, porsaardhur also means newcomer. dukem seem, appear is a middle verb, from the root duk-, which has no clear etymology. Do ti dukej is 3sg. conditional, literally it would appear to him (ti). ky, fem. kjo are demonstratives, probably from something like *kwo-so, *kwo-seh2, if *s can be lost in juncture (Huld). *kwo- is, of course, the interrogative-relative stem seen in L quod. koincidence e trishtuar a sad coincidence. In Albanian, adjectives and nouns are connected by means of an inflected particle, i (m.) viz. e (f.). Of course, trishtuar is a loan-word from Latin (tristis). ai he is the masculine of ajo (cf. above). po vinte is the 3sg. progressive imperfect of vij come, which is a Latin loan-word (venio >> Alb. vij). n in is the same proposition as G en, L in, and OIr. in, PIE *en. Shqiperi as the name for Albania, and Shqiptar for Albanian, are attested only from 14th century onward. The earlier ethnonym, Arbresh, is still preserved in the Albanian enclaves of Southern Italy and in Croatia (Arbanasi). Both ethnonyms are etymologically unclear, however. nga from. Etym? nj one, also indefinite article (indeffinite articles usually come from the numeral one). Though details are unclear to me, it is certain that nj is the same word as L unus, OIr. in, PIE *oynos (or *H3eynos). shtet country must be from Italian stato, rather than from Latin, because L status does not mean state. pr to is a preposition, comparable to L per, G per, etc. trheqjen is a verbal noun from the verb trheq to pull, take back, containing the prefix tr-, and the root heq pull, draw. Alb. heq (Salamis helk) is usually compared to G hlko draw, pull, L sulcus furrow, but this etymology has serious flaws: PIE *scannot give Alb. h-, and positing a laryngeal by-form *Hwolkeyo (Hamp) creates more problems than it solves. eshtr bone is a pluralized form of asht bone, a word with impeccable PIE etymology, cf. Hitt. hatai, OInd. asthi bone, G oston, L os, ossis. The proto-form was probably, *H2ostH1, though some prefer *H3estH1. ushtar soldier is related to ushtri army, contained in the title of the novel. Both are ultimately from L exercitus army. Eshtrave t ushtareve are both in the genitive plural case. The ending -ve is not related, as some people claim, to the PIE D-Abl. pl. ending *-bhyos vrar killed is related to the verb vras kill. These words have no etymology. ktu here contains the PIE interrogative-relative pronominal stem *kwo-, and a particle, perhaps *tu (a similar particle, tu, is attested in OInd.

36

luft war is from Vulgar Latin luctu. The change ct > pt is also characteristic of Balkan Romance. fund bottom, end, from L fundus. botror worlds is an adjective from bot world. The etymology of this word suggested by Meyer (who relates it to L betere to go) is anything but persuasive. Bisedim conversation, negotiation is another loanword from Slavic, cf. Croat. (arch.) besjeda talk, speech. midis between is the same word as midis middle, center. Since PIE *medhyos middle (OInd. madhya-) gave Alb. mjez middle, the history of midis is unclear. There is also Alb. mjet middle, as if from PIE *medho-, with dialectal final devoicing. dy two can be derived directly from PIE *duwoH (OInd. d(u)v, L duo, etc.). See the Phonology section. qeveri government. Etymology? filluar is the participle of filloj begin, and kishin filluar is 3pl. plusquamperfect I. pranver spring is from Romance (VLatin?) primavera por means but, and has no clear etymology. perfundimtar final, adjective containing fund end, bottom, from L fundus. u nnshkruam was signed is passive aorist of nnshkruej to sign, which contains the Latin loan-word shkruej < scribo. Gusht August is obviously from L augustus (mensis). taman just is a Turkish loan-word attested in most languages of the Balkans, cf. Serb. taman just, Tur. tamam. koh time, season is certainly the same word as OCS as moment, but the protoform is disputed: either *keso-, or, more probably, *kesko- (see above). kur when is either from L qua hora (it has to be an allegro-form, since the vocalism does not match), or from something like *kwur- (cf. Lith. kur where). filluan is 3sg. pres. of filloj begin (see above). vranesir is a rare word. It can be translated as gloomy day. i par first should be derived from PIE *prHwo- first (cf. OInd. purva-, Croat. prvi). Another form, *prHmo-, is found in Baltic (Lith. prmas) and Germanic (OEng. forma). pra means therefore, in consequence. As usual with such words, the etymology is obscure. ishte is 3 sg. imperfect of sht < *H1es- to be, but its vocalism is unclear. e tij his, agreeing in gender with the feminine noun koh. The masculine form would be i tij. dinte is 3 sg. imperfect of di to know. This verb is usually compared to OInd. dhiinsight, knowledge. This just might be true, but in such short forms there is always the possibility of chance correspondence.

2. A dialogue between Enver Hoxha and Khrushchev (abridged from E. Hoxha, Vepra, XIX, 372-381) This delightful piece of conversation gives us insight into the mindset of communist dictators. It was published in Enver Hoxhas Collected Works and artistically adapted in a novel by Ismail Kadare. The political background of the meeting is the dispute between Albania and the Soviet Union about the Soviet submarine bases on Albanian territory in the early sixties.

37

Khrushchev (Kh): Nj shoku juaj u ka thn ushtarakve tan se Hrushovi nuk sht marksist. Enver Hoxha (E): Lidhur me shtjen e ushtarakve ne u kemi folur shokve tuaj. interes kemi ne q ushtarakt tan grinden n bazn e Vlors! Kurse ju nxirrni dokumente se nj shoku yn paska thn kshtu e ashtu. Shikojni mir ushtarakt tuaj. Kh: Shoku enver, mos e ngrini zrin! E: Nse do t hiqni bazn, do t bni nj gabim t madh. Ne kemi luftuar edhe pa buk e t zbathur, por asnjher nuk i jemi prkulur askujt. Kh: Nndetset jan tonat. E: Tuajat dhe tonat, ne luftojm per socializmin. Territori i bazs sht yni. Pr nndetset ne kemi marrveshje t firmuara q i njohin t drejta shtetit shqiptar. Un mbroj interesat e vendit tim. Kh: Ju nxeheni, ju m keni pshtyr, me ju nuk mund t bisedosh. E: Ju gjithnj thoni se ne jemi gjaknxeht. Kh: Ju shtrembroni fjalt e mia. Prkthyesi a ei di rusishten? E: Mos u kapni pas prkthyesit, ai e di fort mir rusishtn Englished: Kh: One of your comrades told our soldiers that Khrushchev is no Marxist. E: We have discussed the question of the soldiers with your comrades. Why would it be in our interest that our soldiers should quarrel in the Vlora naval base? While you draw documents that one of our comrades allegedly spoke thus or otherwise. You should watch your soldiers better. Kh: Comrade Enver, do not raise your voice! E: If you pull out with your naval base, you will make a big mistake. We hfought without bread and we were barefoot, but we never bowed to anyone. Kh: The submarines are ours. E: Yours and ours, we are fighting for socialism. The territory of the base is ours. We have signed treaties about the submarines that recognize the rights of the Albanian state. I defend the interests of my country. Kh: You are getting excited, you spat on me, one cannot talk with you. E: You always say we have warm blood. Kh: You distort my words. Does the interpreter speak Russian? Do not take it on the interpreter, he speaks Russian very well. Etymological comments: shoku comrade (def. Nom. sg.) << L socius ushtarak soldier, from the same root as ushtri army << L exercitus lidhur is a participle of lidh bind < PIE *leyg- (L ligo). shtjen Acc. sg. def. of shtje question << L quaestio paska thn allegedly said (admirative, 3 sg. of thom say)

38

zrin Acc. sg. def. of z, zri voice, from *ghwono- (OCS zvon sound). kemi luftuar is 1pl. perfect of luftoj fight << L luctare buk bread is probably from L bucca mouth, bite , cf. Middle Greek bokka biscuit, Romanian buc food. nndetset submarines (Npl. def.) is a compound containing det sea (of disputed origin, perhaps from *dhewbeto-, cf. OE deop deep). marrveshje understandings is compounded from marr take and vesh ear (< *h2ows-, OCS uxo ear). firmuara Npl. of the participle of firmoj sign, from Italian firmare njohin 3pl. ipf. of njoh recognize < *gneh3-sk- (L co-gnsco). drejta rights is Acc. pl. of drejt << L directus shtetit Gsg. of shtet state from Italian stato. vendit Gsg. of vend place, region, country is derived from the ve to put, of unknown origin (perhaps related to Goth. winja pasture). t bisedosh (2sg. subj. pres. of bisedoj talk), from Slavic (OCS besda talk). gjithnj always is compounded of gjith all and nj one; the etymology of gjith is disputed: it is usually segmented into gji- and the suffix th-, and the element gji- is derived from PIE *sem- one (L semel once), but the phonetic development is doubtful. gjaknxeht warm-blooded is a compound containing gjak blood < *h2esr blood (L sanguis) fjalt (Acc. pl. def.) from fjal word << L fabella tale. fort strongly, very << L forte

3. A passage from Buzuku (1555) in Old Gheg (Luke 16, 1-4) in transcription. 1. nd ate mot: tha Jezu dishipujet vet: Ish nj nier fort i begat, qi kish nj spenxatuor; e k kle shtm nd t madhe desgracje prpara t, ai tue i faruom gjith gjan e t. 2. E grishi, e i tha: Qish ansht kta qi gjegjem n teje ? A m arsye n gjithsei k prdm; prse ti m nuk t mundnjsh kha paret m me prdm. 3. E tha spenxatori nd vethen: Qish t banj, se em zot m muor bukn teme e m spenxatuor nuk t jm? me mihun nuk mundinj, me vot pr dyer kam dhun. 4. d qish t banj, qysh ai m mart kte buk, prap t i jes nd shtep. Englished: 1 At that time Jesus said to his disciples, There was a very rich man who had a steward, and he [the steward] was put into great dishonour before him, (they said that) he was wasting all his goods. 2 He called him and said to him, What is this that I hear about you? You must hand me over the proof of all, for you can no longer be steward. 3 And the steward said to himself, What shall I do, since my master took my bread and I am not to be a steward anymore? I am not strong enough to dig, and I am ashamed to go begging. 4 I know what to do, since he took this bread from me, so that still I may stay in their houses. Etymological notes:

39

mot 'time' < PIE *meh1-to-, from the root *meh1- 'to measure' (Skr. mti 'measures', OCS mra 'measure'). dishipujet 'disciples' < L discipuli vet 'own', reflexive particle < *h2ewto- (G auts) ) nier 'man' (Std. njeri) < PIE *h2ner- 'man' (G an r fort 'very' < L forte begat 'rich' < Slavic (OCS bogat) kish 'had', ipf. 3sg. of ka 'have' (< *kap-mi, L capio, etc.) spenxatuor < Italian dispensatore (lit. 'the dispenser') kle shtm 'was put'; kle is the Old Albanian form of Std. Alb. qe 'was' (aor. 3. sg. of jem 'be'), and shtm is the Gheg participle of shtie 'throw'. Etym? madhe fem. of madh 'big' < PIE *meg'h2- 'big' (L magnus etc.) desgracje < Italian disgrazia gjith 'all' < ? (perhaps related to PIE *sem- 'one', but this is disputed, see above). gjan pl. of gj 'thing' (Tosk gj, gjri). Etymology unknown (the connection with L is 'soup', found in some handbooks, is too speculative). grishi aor. 3 sg. of grish 'call upon'. Etymology unknown (sometimes speculatively connected with Lith. gar sas 'voice, sound'). tha aor. 3 sg. of thom 'say' < PIE *k'ens- (L censeo, etc.) ansht Gheg 3 sg. pres. of jam 'be' (Std. sht) gjegjem 'I hear' (Std. dgjoj), Etym? arsye 'proof, reason' (Std. arsye) << L rationem 'I', Gheg form of Std. un < *h2ew-nu (?), where *h2ew- is the reflexive particle (G auts) t banj 'do' (1sg. pres. subj.), Std. bj, probably from PIE *bheh2- 'say' (L fari) zot 'lord, master', without etymology. Some derive it from *dyew- '(sky-)god' (G Zes), but this is too far-fetched. muor aor. 3sg. of marr 'take' me mihun Gheg infinitive of mih 'dig'. Etym? mundinj 1sg. pres. of mund 'can' (impersonal in Std.) me vot Gheg infinitive of votoj 'vote, vow, implore'; in the phrase vot per dyer 'beg for mercy' (lit. 'ask by the doors'), where dyer is the pl. of der 'door' < *dhwr- (L fores, OCS dvr). kam 1sg. pres. 'have' dhun 'damage', shame'. Etym? mart opt. of marr 'take'. Etym? buk, def. Acc. sg. bukn() 'bread' << L bucca (see above) di 1sg. pres. of di 'know', perhaps from *dhiH- (Skr. dh- 'insight') t'...jes 1sg. subj. of jes 'stand, remain'; perhaps related to jetoj 'live' << L aetas 'life' shtep 'house' << L hospitium

40

4. Cinderella (Fragment of a folk story in the Gheg dialect from Arbanasi in Croatia, from Tagliavini 1937). Br-i i krl-it i ko puvt qish ko me ken shoto btit, ma at son of king.gen.sg. them has asked what has to be under barrel.Asg., but they.fem. i kan than da nu(k) ko krgj, ali gjl-i ko zan me kantt him have.3pl. said.3pl. that not has nothing, but the.rooster has started to sing e br-i i kral-it prap ko puvt qish munje me kn shoto but-it and the.son the.king-gen.sg. again has asked what could to be under the.barrel.gen.sg. e at prap i kan thn da nu(k) ko kurgj. Ma t trtn er and they.fem. again him have said that not has nothing. But the third time kur gjl-i ko kantt thi prn er: Kikirku when the.rooster has sung as first time: Cock-a-doodle-doo lipa cura pot koritu... [Croatian] beautiful girl under the trough ai i ko bo me iq but-in. Apna he them has made to raise the.barell-acc.sg. Ass soon as bt-in e kan iq the.barrel-acc.sg. it have.3pl. raised gji ka(n) mbt all have.3pl. remained

t qudtum e mo shum pe gji maqi-a e b-at t 'thi kur e the.pl astonished and more a.lot of all the.stepmother and the.girl.pl. of her when her kan po Kucocnera t vsht gji n or inveqe at kan have.3pl. seen Cinderella the.fem. dressed all in gold on.the.contrary they.fem have.3pl. thought menda m'e po gji asht t shtrancte. Bir-i i kral-i(t) ko vt me thought to.her see all so the.fem.shabby. The.son the.king-gen.sg. has gone to ja mt kapcn; apna ja ko ti n kamb ko po her measure the.shoe-acc.sg. as.soon.as her has put on foot has seen her da bash sht e thoj that just was hers i bkr e ko mar e ko kunr; math pak mot kt b-at he nicely her has taken and has married; after little time those.fem daughter-nom.pl. pe t dyte grue kan art n jat me lyp e of the second wife have.3pl. come to her to beg(mercy) and athoi jo bo mlo math gji at qish i kan bo e ju ko lan

41

to them it made merciful after all that what him have done and them has given shum shola. a.lot money NOTE: This dialect has undergone strong influences from both Croatian and Italian (the Venetian dialect). Tagliavinis informants in 1937 were all trilingual. This can be seen in many loanwords from both languages, as well as in some syntactic constructions (e.g. the use of Croatian subordinator da, the existence of Romance-type analytic causative with bj do + infinitive, etc.). It also shows some special phonological developments, e.g. the change of Alb. q to [t], gj to [d], the loss of Alb. h and the Gheg nasal vowels, the change of a to o, the sporadic change of th to s and of dh to l, the merger of Alb. r and rr (there is only ordinary r in Arbanasi), the loss of some final consonants, etc. Etymological comments: apena as soon as is, of course, from Italian appena. art come (Standard Alb. ardhur) is a participle derived from the PIE root *h1ergh- go (OIr. eirgg go!, G rkhomai). bash just is from Croat. ba bir -ison, as well as bij daughter have been derived from a PIE root *bhey- shoot, spring (Gr. phty sprout, Arm. bir stick, club, but I doubt this etymology. The connection with L flius son is impossible, as this is from PIE *dheh1-i- to suck. bo do, make (Gheg. ba , Standard bj, participle br) is derived from Proto-Alb. j *bany make appear, from PIE *bheh2- shine, appear (G phan, Skr. bh-). but-i barrel is from Venetian (cf. Italian botte). da that is the Croatian subordinator (da). er time (Standard Alb. her) is usually derived from Lat. hra, but the initial h- is st unexplained (h- was lost in Latin pronunciation in the 1 century BC). gjel-i rooster is from L gallus. The vocalism is due to singularization. gji all (Standard Alb. ghith) has been derived from PIE *sih2-ko-, where *sih2- is the feminine form of the demonstrative *so- this (cf. Gr. a one). As speculative as any Albanian etymology. w grua wife, woman, pl. gra is usually derived from the PIE word for woman, *g enh2 (G gyn, OCS ena, etc.), with irregular change of *gn- > *gr-. inveqe on the contrary is from Italian (invece). iq raise (Standard Alb. heq) is usually derived from PIE *selk- (G hlk draw). kamb foot (Standard Alb. kmb) is from Vulgar Latin camba leg (cf. It. gamba). kantat sung is from Italian (cantare), probably through Croatian (cf. Croat. dial. kantat sing). kral king is from Croatian (kralj). w kur when has been compared to Lith. kur where (PIE *k ur?). mar (Standard Alb. marr) take is usually connected with G mr hand (which is a dubious word), L mnus hand, PIE *meh2- (cf. also OCS manti give a sign, give a wink). mat measure is probably from PIE *meh1- measure (OCS mra, L mtior). me with, occuring in the Gheg infinitive (preceding the participle) is probably related to OE mi with and Skr. smat with, but details are unclear. mendua thought (Standard Alb. mendoj measure) is a denominative verb built from mend thought, which is a loan from L mente42

milo merciful is from Croat. mio, mila dear pak a little is from L paucus par first is from PIE *prh3- first (cf. G pr tos< *prh3to-). prap again has been derived from the preposition pr and some particle (? *ap, para before), but this is not very helpful. po seen (Gheg. pa, Tosk. pash) is the suppletive aorist stem to shoh see. The often-quoted derivation from PIE *spek- see (G skptomai, L spicio) is difficult, if not impossible. quditum astonished is from Croatian (uditi astonish). shola money is from Venetian (cf. Italian soldi). shoto under is from Venetian (cf. Italian sotto). shtrancat shabby is from Venetian (cf. Italian stracciato) shum a lot is from L summum the highest than (Standard Alb. than) is the participle of thom say, which is generally derived from PIE *kens- praise (L cense, etc.). zan (Standard Alb. z, Gheg. za) catch, grasp, start etymologically obscure. It has wh been derived from PIE *g en- strike (Skr. hanti, Hitt. kuenzi, assuming the operation of Pedersens law).

43

REFERENCES
Boretzky, N. Der trkische Einfluss auf das Albanische, I-II, Wiesbaden 1975-6. Camaj, M. Albanische Wortbildung, Wiesbaden 1966. Camaj, M. La parlata albanese di Greci in provincia di Avellino, Firenze 1971. abej, E. Hyrje n historin e gjuhs shqipe, Prishtin 1970. abej, E. ber einige Lautregeln des Albanischen, Die Sprache 18/1972: 132-154. abej, E. Studime gjuhsore I-VIII, Prishtin 1988. Demiraj, B. Albanische Etymologien, Amsterdam-Atlanta 1997. Demiraj, Sj. Gramatik historike e gjuhs shqipe, Tiran 1986. Demiraj, Sh. Gjuha shqipe dhe historia e saj, Tiran 1988. Desnickaja, A. V. Albanskij jazyk i ego dialekty, Leningrad 1968. Elsie, R. Early Albania. A reader of historical texts 11th-17th centuries, Wiesbaden 2003. Fiedler, W. Das albanische Verbalsystem in der Sprache des Gjon Buzuku (1555), Prishtin 2004. Gjinari, J. Dialektet e gjuhs shqipe, Tiran 1989. Gjinari, Jorgji, Bahri Beci, Gjovalin Shkurtaj, Xh. Gosturani, Atlasi Dialektologjik i Gjuhs Shqipe, I-II, Akademia e Shkencave e Shqipris / Universit degli Studi di Napoli lOrientale, 2007, 2008. Haarmann, H. Der lateinische Lehnwortschatz im Albanischen, Hamburg 1972. Hamp, E. P. Evidence in Albanian, in: Evidence for Laryngeals, Den Haag 1965: 123141. Mamp, E. P. Albanian, in: J. Gvozdanovi (ed.) Indo-European Numerals, Berlin 1992: 835-922. Hock, W. Zur Vorgeschichte des albanischen Lautsystems, in: Sprachkontakt und Sprachwandel, ed. by G. Meiser & O. Hackstein, Wiesbaden 2005: 261-274. Huld, M. Basic Albanian Etymologies, Columbus 1984. Ivanov, V. V. Slavjanskij, baltijskij I rannebalkanskij glagol. Indoevropejskie istoky, Moskva 1981. Jokl, N. Linguistisch-kulturhistorische Untersuchungen aus dem Bereiche des Albanischen, Berlin 1923. Klingenschmitt, G. Das Albanische als Glied der idg. Sprachfamilie, in: J. E. Rasmussen (ed.) In honorem Holger Pedersen, Wiesbaden 1994: 221-233. Kortlandt, F. Albanian and Armenian, KZ 94/1984: 243-251. Lafe, Genc Der italienische Einfluss auf das Albanische. Erster Teil Allgemeine Errterungen PontoBaltica 89, 1998-9: 47172. Matzinger, Joachim, Der Altalbanische Text Mbsuame e Kreshter (Dottrina cristiana) des Lek Matrnga von 1592. Eine Einfhrung in die albanische Sprachwissenschaft. Dettelbach: J.H. Rll, 2006. Matzinger, J. Messapisch und Albanisch, International Journal of Diachronic Linguistics and Linguistic Reconstruction 2/2005: 29-54. Meyer, G. Etymologisches Wrterbuch der albanesischen Sprache, Strassburg 1891. lberg, H. M. Untersuchungen zum indogermanischen Wortschatz des Albanischen und zur diachronen Phonologie, auf Grund des Vokalsystems, Innsbruck 1972. lberg, H. M. Einige berlegungen zur Laryngaltheorie (an Hand des Albanischen), KZ 86/1972: 121-135. Orel, V. Albanian Etymological Dictionary, Leiden 1998. Orel, V. A Concise Historical Grammar of the Albanian Language, Leiden 2000. Pellegrini, G-B. Avviamento alla linguistica albanese, Palermo 1995.

44

Tagliavini, C. Lalbanese di Dalmazia, Firenze 1937. de Vaan, M. PIE *e in Albanian, Die Sprache 44/2004: 70-85. Vermeer, W. The prehistory of the Albanian vowel system: A preliminary exploration, in: Evidence and Counter-Evidence. Fs. Kortlandt, Vol I. Amsterdam 2008: 591608. Ylli, Xh. Das Slavische Lehngut im Albanischen. 1. Lehnwrter, Mnchen 1997.

45

ABBREVIATIONS
A) Languages Alb. = Albanian Arm. = Armenian Croat. = Croatian G = Greek Goth. = Gothic Hitt. = Hittite It. = Italian L = Latin Lith. = Lithuanian OCS = Old Church Slavonic OEng. = Old English OHG = Old High German OIr. = Old Irish ON = Old Norse PIE = Proto-Indo-European Rom. = Romanian Russ. = Russian Serb. = Serbian Skr. = Sanskrit Std. = Standard (Albanian) ToA = Tocharian A ToB = Tocharian B Tur. = Turkish W = Welsh B) Grammatical terms A = Accusative Abl. = Ablative aor. = aorist D = Dative def. = definite f., fem. = feminine fut. = future G = Genitive indef. = indefinite ipf. = imperfect m. = masculine N = Nominative n., neut. = neuter neg. = negation opt. = optative part. = particle pl. = plural pres. = present sg. = singular

46

< = is derived from << = is borrowed from * = unattested form

47